Can students use AI and still become critical thinkers? - 26 April 2026

In the last week I asked some college professors how they dealt with AI cheating.

Stephen Kotkin. Nation's pre-eminent Russia expert. Hoover Institution Fellow. Stanford University. Look. They're going to use AI after they graduate. Better to teach them risks of substituting AI for critical thinking. Allow students to use AI in such a way so as to advance critical thinking skills. I think two thirds of the students get it, and they'll be better off for it compared not using AI at all,

John Cochrane. The Grumpy Economist. University of Chicago. Hoover Institution Senior Fellow, Stanford University. Cochrane said he wasn't teaching classes at present. But if he were teaching, he'd have the students take exams in a digital free room using pen and "old fashioned" blue books.

Brigitte Madrion. Dean, Brigham Young University School of Business. You can't stop students from using AI. You must teach them to use AI in such a way that does not diminish critical thinking skills. Taught the right way, AI can "expand the agency" of a student user.

A couple of observations.

For the last ten years or so, I have done book reports on the books I read. Today, I can go to Grok and pull off in seconds a book report as comprehensive as any report I could do myself in a couple of hours. The temptation is great to substitute, or adapt, Grok for my book report. I mean completing a report in fifteen minutes as compared to two hours? Cummon!

Our ROMEO group, LSDM, has speakers. We have had over nine-hundred speakers in the last fifteen years. For all speakers we have captured notes for those talks which we archive on our website: La Société Deux Magots - Home  

With the advent of Zoom AI Companion, a virtual note taking synthesis/summary process, I no longer take the speaker notes myself. I let Zoom AI Companion do it. Of course I review the AI Companion output. Early on in using AI Companion, I would have to make quite a few edits. The note taking synthesis feature often missed the gender of the speaker and would omit key points made by the speaker that I thought were germane. I would edit the output accordingly. Now, two years later, I make very few edits to an AI Companion speaker note. The Zoom AI Companion now produces a near perfect synthesis of what the speaker was trying to get across, even without gender mistakes. In fact, for a less than well-organized talk, the AI synthesis will now outline what the speaker probably wanted to say had he/she organized the talk better!

It really doesn't matter for me anymore. I'm eighty years old. Let AI do the work. But I suspect, in using AI to help with book reports and by letting AI do, sans serious edit by me, speaker notes for LSDM, my own critical thinking is taking somewhat of a dive.

I applaud Stephen Kotkin's and Brigitte Madrion's efforts to build critical thinking capacity in their students while allowing use of AI. But considering my own experience of getting a top product with critical thinking laziness, were I a teacher, I'd do a hybrid of the aforementioned AI teaching strategies: Attempt to teach critical thinking while letting students use AI to "expand agency," but end the class with a final exam taken in an analogue room with pen and blue book... just to make sure.