Has Anyone Run Their OWN Writing Through a Detector? You’d Be Surprised.
I asked this as a one-liner in another thread but it deserves its own discussion.
Has anyone systematically run their own teacher-written content through AI detection tools? not a student essay – YOUR writing. Your syllabus, your assignment handouts, your lesson plans, your emailed responses to parents.
I did this after getting suspicious. Results:
– My syllabus: 31% GPTZero
– A lesson plan I wrote last week: 44% GPTZero
– An email I sent to a student explaining feedback: 28%
– A handout I wrote in 2019: 14%
The more formal and structured my writing, the higher the score. which makes sense – i write clearly and organize my thoughts consistently. so does AI. the tools can’t tell the difference.
what this tells me: ANY teacher applying these tools to formal student writing is working with a tool that would also flag their own professional output at rates they’d consider unacceptable.
try it.
5 Replies
Join the discussion.
Log In to Reply
ran my department handbook section through Turnitin last month. 44% AI. co-written with a colleague over three weeks of meetings, every word deliberate. If the tool would flag my own professional writing at that rate, I have no business using it to make judgments about student work. The self-test exercise every teacher should do before using these tools.
Liam's implication is right - if we got a 38% score and can explain exactly how and why we wrote what we wrote, imagine what score a nervous student with a genuinely unusual writing style gets with no ability to explain themselves in an appeal. the human context is everything and the tool has none of it.
ran my last annual report through a detector. 71%. it had been reviewed by three administrators before I submitted it. I write the same way I've written for fifteen years. take that for whatever it's worth.
I got 44% on a lesson plan I wrote from scratch in 20 minutes with zero AI. something about my writing style apparently reads as very AI. this is exactly why I stopped running these on student work.
my own writing got flagged at 38% on GPTZero. I've been teaching for 12 years. It's my own voice. The tool just can't distinguish between a formal human writer and AI, because AI learned to write formally from formal human writers. the irony is complete.