What I'm reading: AI's effects on education
I'm down a rabbit hole at the moment, exploring what role AI should have in higher education. The fun thing is that, like EdTech, we have to figure out what to do with this technology now, but we probably won't know the results of our decisions for several years (possibly a decade).
Currently, I'm of the opinion that AI can be used in beneficial ways, but it often isn't, and the temptation to use it in less beneficial ways may push it toward being a net negative for educational development. Perhaps with the right harnesses in an educational setting it might be possible to curb the downsides enough. But we have to be very careful about how this is deployed and incorporated into classrooms.
The challenge, as outlined in the video below, is that I think friction and repeated practice are important, but they aren't fun, and GenAI makes it easier than ever to skip those things because it doesn't feel like you need them to be productive. But the point of writing an essay was never the product you created. It was what you learned in the process of creation. The point of creating a tic tac toe game in Python was not that we didn't have a million of those. It was how the process of creating the game ingrained programming concepts into your understanding.
There are ways that we can use GenAI to improve education, but in doing so we (and especially students) need to avoid using it in ways that harm.
Derek Veritasium (Muller) - What we get wrong about AI and Learning
Good pedagogical overview for what education does to our brains.
AI assistance reduces persistence and hurts independent performance
Liu, G., Christian, B., Dumbalska, T., Bakker, M. A., & Dubey, R. (2026). AI assistance reduces persistence and hurts independent performance (arXiv:2604.04721). arXiv. https://doi.org/10.48550/arXiv.2604.04721
This one got a lot of hype when it came out a few weeks ago. Not a perfect paper, but it shows an interesting phenomenon - people who use AI get used to having AI, and become less willing to cognitively exert themselves. Study 3 in the paper (the reading comprehension problems) is particularly concerning to me - even if AI doesn't actually make you better, you use it and lose persistence.
Thinking about extending this paper a bit to understand how motivations change the outcome.
Impact of ChatGPT on math creativity
Liu, Z., Li, G., Zuo, H., Lu, Y., & Feng, J. (2026). Exploring the impact of ChatGPT on college students’ mathematical creativity: A quasi-experimental study. Australasian Journal of Educational Technology, 42(2), 65–85.
Shows that students using ChatGPT can come up with more creative responses to math questions. These math questions require creativity (it's not just new creative ways to solve for x). Like many studies that show that ChatGPT improves performance, what it really shows is that ChatGPT can do those problems fairly well, and students can use ChatGPT to coax more creative answers. And like most of those studies, it doesn't do anything to address if the students actually learn anything.
I can use ChatGPT to generate an essay about Napoleon, but that doesn't mean I learned anything about Napoleon.
From tool use to evaluative judgment
Hall, M., & Hawkins, D. (2026). From Tool Use to Evaluative Judgment: A Readiness Framework for Generative AI in Higher Education (SSRN Scholarly Paper No. 6598018). Social Science Research Network. https://doi.org/10.2139/ssrn.6598018
A framework from a friend at UNOmaha for understanding how courses and students can be prepared for GenAI usage. It always gives me hope to read about students who mindfully and intentionally approach their AI usage. I get caught up in horror stories and cautionary tales, and sometimes forget that most students actually care about their educations.
A student's complete guide to thinking better with AI
Keith, M. (2026). A Student’s Complete Guide to Thinking Better with Artificial Intelligence. https://aimodes.ai/Student-Guide-to-AI-Assisted-Learning.docx
I'm not sure where this one is publicly available, but my colleague Mark Keith has a very detailed rubric for AI usage. Great recommendations for how to use AI to practice your skills and deepen your understanding of topics. You have to move beyond asking it for answers. Use GenAI to provide instant, individual feedback, something we typically struggle to provide in a classroom.