Higher Ed Musings 8.
Attention Must Be Paid
Whatever your views on University of Austin, the brand new “revolutionary institution” founded by former St. John’s College President Pano Kanelos and a group of well degreed iconoclasts that welcomed its first students this past week, UATX, as it is called, should be praised for its embrace of the idea that a carefully selected human teacher is the interface central to the education endeavor.
Compare UATX, with its list of names that you either admire or not, depending on your political stance — Niall Ferguson, Ayaan Hirsi Ali, Larry Summers, Nadine Strossen, Glenn Loury, Deirdre McCloskey — and its commitment to designing a curriculum with and around its faculty, with the recent Civics Alliance/NAS gen-ed curriculum initiative, a long list of Western Civ courses that is silent on who should teach the courses, all but proclaiming: “the teacher doesn’t matter.” Great texts, the assumption is, “teach themselves.” All over the country, legislative interference with public schools and universities is a matter of the stuff to be taught, not the quality or expertise of anyone teaching it.
The question of who is teaching, of the value of great teaching, matters more than ever in the AI/LLM era where every day new startups are launching new forms of personal instruction models that celebrate the fact no human teacher is needed.
Until technology is invented that transports information from a source directly into the human brain, a learner is going to need to pay attention to an information source in order to learn. The paying attention is the key. The better a teacher is at getting a learner to pay attention, the better the instruction. Right now, LLMs are fun. Those of us who pay attention to them pay attention to them. Does every user? Depends on what you bring to the table.
As Tyler Cowen smartly observed, “the marginal product of LLMs is when they are interacting with well-prepared, intricately cooperating humans at their peak, not when you pose them random queries for fun." The more expert you are, the more you will learn from LLMs.
My favorite undergrad philosophy professor, Michael Hooker, used to cite Jean Piaget who observed that if you work to become an expert in one thing, however narrow, you gain the ability to be an expert in other things — you’ve trained your brain to organize knowledge effectively. Piaget was a world's expert in the classification of freshwater mollusks at age fifteen.
The best scholars are those who seek to be a world’s expert in something, who have a single-minded focus on that thing, and strive to learn everything about it and change how we see it. In the case of Piaget, the topic was small and circumscribed. It is impossible to be an expert in “religion” or “sports,” for example, but you can become a world’s expert on a particular Medieval saint or plastic-eating fungi and tell the world why that’s important.
Of course there are generalists too. Charles Darwin had almost no specific topic expertise when he boarded The Beagleas the ship’s naturalist at age 23, though he had a great deal of practical expertise in the art of collecting things like beetles and saltwater mollusks. If he had been an expert in botany he would have focused on plants on his famous voyage. If zoology, on animals. As an amateur he focused on everything, honing his real expertise for five years—collecting—and in the process, categorizing and classifying, which led to his great theories.
But Darwin had great teachers (notably John Henslow). And Darwin’s great champion, Thomas Huxley, was an expert in marine invertebrates, which was part of why he recognized Darwin’s genius. The best way to recognize an expert is to be one. Michael Hooker was an expert in the philosophy of education as well as a brilliant teacher to whom I paid attention, when I slept through so many other courses. The best way to become an expert is to learn from experts who make you pay attention.
Long before Covid, MOOCs and hundred of internet courses threw tens of thousands of teaching hours open to public view, in credit-bearing online courses and informal video recordings of lectures. Much of what one sees is boring and horrible. Take a look. When you pay attention, what do you pay attention to? Is it the content? The teacher? The graphics?
(There was a fascinating study a few years ago that suggested that students like entertaining lecturers that make them think that they’re learning. I’d put most TED Talks in this category – everyone goes home full, thinking they’ve learned a lot, when in fact when you are seen and engaged with by the teacher you are more likely to retain the information.)
Even in this AI era, gatekeeping for the best teaching, by great expert professors, has not yet been disrupted. First, there aren’t standard long lists of great professors across the country or around the world. Second, when you identify one, you usually have to be enrolled in the school where they teach. Maybe holograms someday but not yet. Great teachers are still only known only by a select few. Charles Williamson, professor of engineering at Cornell, is an expert in many fields as well as a teacher and mentor to decades of engineering students. I know him because my son, now an engineer at Skydio, studied with him. But the gatekeeping system of college admission limits who can study with him.
Too few people inside and outside the education sector talk about what makes a great teacher and what makes great teaching—what makes people pay attention. AI has not yet helped this problem. Most of the Ed-tech startups I see focus on information delivery, not teaching, not the interface between the expert source and the learner.
I’d like to see an AI future that focuses on what makes a great teaching interface, what makes users pay attention, and will make more people better learners. Until then, UATX is the only startup I see putting the human interface first. Given my last post I of course quibble at Bari Weiss’s identity-making line “Welcome to the class of 2028.”