Using ChatGPT to differentiate reading levels for the same lesson, game changer or crutch?
This year I started using ChatGPT to generate three reading level versions of the same historical source document for my grade 9 class, so students access the same content at a level that matches their reading ability. Preliminary comprehension quiz scores are up almost 12 percent compared to last year when everyone read the identical text. I want to be careful about overselling this though, the sample size is small, one term of one class, and I have not controlled for the fact that this is also a stronger cohort overall this year than last year. Still, the differentiation piece feels genuinely useful in a way a lot of the other ai classroom use cases have not for me personally. Curious if anyone else has used ai this way and what pitfalls to watch for before I lean on it more heavily next term.
5 Replies
Join the discussion.
Log In to Reply
@Ethan Roy fair, and honestly i said as much in my post, i dont have a clean control group this term. planning to run it properly next semester with two sections of similar ability doing the standard text versus the differentiated versions.
YES to differentiation use cases in general, this is one of the areas where ai actually seems to help the kids who need the most support instead of just making life easier for the strongest students. tried something similar for a novel study last spring.
worth watching for is if the simplified versions accidentally strip out content students need for the actual assessment later. differentiated reading is only useful if the assessment doesnt assume everyone read the hardest version.
@Rachel Dubois thats a real risk i hadnt fully thought through, need to double check my quiz questions arent inadvertently favouring details only in the hardest version.
i'm not sure the comparison holds without a control group reading the original text this term too. 12 percent could easily be the stronger cohort, not the differentiation.