Education Links, 5/27/2025

Clay Shirky on AI and student learning; Michael Strong on teaching ethics; Naomi Schaefer Riley and Jim Piereson on financial aid; Tyler Cowen on the plight of universities regarding AI

Clay Shirky writes,

This preference for the feeling of fluency over desirable difficulties was identified long before generative AI. It’s why students regularly report they learn more from well-delivered lectures than from active learning, even though we know from many studies that the opposite is true. One recent paper was evocatively titled “Measuring Active Learning Versus the Feeling of Learning.” Another concludes that instructor fluency increases perceptions of learning without increasing actual learning.

On this last point, I observed that over the years that I taught AP statistics, the better I got at explaining, the worse I got at teaching. It was better for students if I stumbled, back-tracked, or used a more challenging way to demonstrate a proposition than if I did so quickly and efficiently. I see this also in Israeli dancing, where the more mistakes that the teacher makes and has to correct, the better is my memory for the dance.

Learning requires work. When the teacher stumbles, that can force the student to work harder. This relates to AI, because AI can allow students to get away with less work. This appeals to students, but it does not help them.

Michael Strong writes,

I don’t believe that any simple dogmatic preaching of morality will become influential going forward. At the same time, I believe that creating a social culture in which the rational discussion of moral principles is the norm, starting in adolescence, is crucial to individual and social well-being going forward. This is especially the case as the shadows on the wall become ever more addicting, as AI will appeal to our short-term impluses more seductively every time we are online (and insidious ways no doubt encroaching on the rest of our lives as well).

To that extent, I’m unapologetically in favor of educators dedicating themselves to turning young people’s minds away from the shadows in the cave, away from the temptations of sensual life and impulsivity, and towards the light. And then encouraging them to do so for others across the length of their lives.

Naomi Schaefer Riley and Jim Piereson write,

Graduate education has to go. This is not a judgment about which parts of higher education produce the most economic impact or equip people for jobs. We want people to be doctors and lawyers and research scientists, but people who have completed a four-year degree and can get into these graduate schools will have a clear stream of income after graduation and should be able to take out private loans against that future earning potential, either from banks or employers.

According to the College Board’s “Trends in College Pricing and Student Aid 2024” report, total financial aid for graduate students grew enormously between 2003 ($39.4 billion) and 2023 ($66.6 billion). During that period, by far the largest source of aid every year was federal loans, topping at 68% in 2009 before falling to 61% in 2023.

Tyler Cowen writes,

Would it be so crazy to put o3, or some other advanced AI model, on the dissertation committee, in lieu of the traditional “outside reader”?

…The most ambitious professors will learn how to use the AI models themselves, and give their students the exact same quality feedback the students will be able to get for themselves. The best among the professors will learn to be inspirational mentors, coaches, and networking connectors. They will very directly help their students get somewhere in the real world. Those are some skills the AIs cannot copy, at least not anytime soon.

He asks an LLM to come up with ideas for reconfiguring higher ed around AI, and he links to the response he got. The response is good, but is it as good as my network university idea?

substacks referenced above:

@

Everyone's Using AI To Cheat at School. That's a Good Thing.

@

Leading Children Towards the Light