How is everyone handling ChatGPT on math homework this year
probably a dumb question but here goes. i teach grade 9 math in a pretty small rural school and this year almost every kid has a phone out doing homework with chatgpt open in another tab. its not like english class where i can tell if the writing sounds off, math is just numbers and steps, so how am i even supposed to know if they understand it or if the ai walked them through it.
some of my strongest students are the ones using it the most, which throws me off even more. i figured it would be the kids who are struggling.
anyone actually found a way to use this instead of fighting it? i dont have the resources some of the bigger schools have, no budget for fancy detection stuff, its literally just me and a whiteboard most days. would love some real suggestions from people who teach math specifically.
2,684 Replies
Join the discussion.
Log In to Reply
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.
not a dumb question at all Raj. this is kind of like the cell phone ban debate honestly, we spent years arguing about the device instead of the behaviour we actually wanted to change. i wonder if the answer with math and ai is similar, we might be fighting the wrong battle. what if the homework itself changed to require the kind of reasoning ai cant just hand over, like explain why a wrong answer is wrong instead of just solving for x.
Brendan's log idea scales pretty well actually, i tried something similar with 90 students across three sections and it wasn't as much extra marking as i expected once i built a quick rubric for it. Andrew's oral checks work great in smaller classes but with my numbers i had to find something i could do async. worth testing both and seeing what fits your class size, Raj.
honestly i went through the same thing last year and ended up flipping my whole approach. now i have kids submit their chatgpt conversation alongside their answer, screenshot or copy paste, doesn't matter. it forces them to actually engage with what the ai said instead of just copying the final number. i tried walter writes ai for a totally different purpose, checking some written explanations kids give for their steps, and it was decent for that but the real win was just requiring the conversation log. tracked accuracy before and after for six weeks, went up almost 15% because kids started actually reading the explanations instead of skipping to the answer.