AI Teachers And the End of Learning

What do LLMs as classroom teachers mean for education?

It seems inevitable that public education will use AI teachers in classroom settings. Likely, each student will have an AI tutor. An in-person teacher and aid may remain to support students with special needs, but the future of public education seems tilted in this direction.

Given the dramatic downturn in student learning outcomes, the current student–teacher ratios in classrooms, the chronic underfunding of education, and the cultural temptation to use AI in every sphere of life, it seems doubtful that any other future will occur.

MacKenzie Price, the CEO of Alpha Schools, has already advocated for such an approach. In her initiative, Alpha Schools, she creates personalized LLMs for each student to maximize their potential. The argument is simple: an LLM can give individualized attention to each student, helping them to learn, in Price’s language, “effectively and efficiently.”

Outside public education, expensive private schools (classical schools, for example) will likely retain in-person teaching as their premier option. Yet some, like Alpha, will also attract parents by advertising the “best LLM possible” for learning.

What is Education?

Here lies the crucial question: what is education? On the Alpha model, education means the accumulation of accurate information, reading ability, mathematics, and similar subjects. On this theory, efficient systems of information delivery describe the end of education.

To be fair, Price also speaks of training children to be learners and creators in a modern economy. But even then, the manner of education remains informational, centred on accumulation and data through LLMs. The kind of creation she envisions is telling: a young girl developing a chatbot that offers safe dating advice. A remarkable act, yes—but one still bound within the plane of technological education and youth-driven knowledge production.

Another vision of education, older and richer, centres on virtue formation. Here the goal is for one human teacher to kindle in students a love for truth, goodness, and beauty, so that they might love what is good and despise what is evil. The aim is to shape a kind of person—deeply human, morally serious—who knows what to do with LLMs, atomic bombs, and penicillin.

The technological model sees education as facts, data, and skills for technological ends. Price herself notes that when her students go to university, they find lectures “boring and inefficient.” Why wouldn’t they? If one can learn more efficiently from an LLM than from a data-driven lecture, the professor becomes unnecessary.

But the virtue model points to a different end: not simply information, but formation. Every progressive already knows this—education is the shaping of opinions, loves, and desires. Modern conservatives, by contrast, have ceded ground by claiming school is about “math and facts, not ideology.” They miss the point.

Granted, education is not ideology, but it also is not merely facts either. Education involves the shaping of our loves so that, by the time we reach the age of reason, we already desire what is good and resist what is evil.

Counting the Costs

Both models are costly. Alpha Schools charge between $25,000 and $75,000 per year. Public education, once one accounts for real estate, administration, and salaries, is also expensive. A virtue-driven approach would also be costly today. So the choice cannot rest on economics alone.

Instead, we must ask: what is lost when education becomes not only informational (as it already largely is) but is also delivered primarily by LLMs rather than by human teachers? One may point to middle-ground approaches like Alpha, which supplements LLM education with outdoor experiences and real human contact. But even here, the central pattern is data, and the classroom still orbits around the machine.

I fear that in asking this question, I am like Ross Douthat asking Noor Siddiqui what might be lost if procreation were separated from sex. The technologist could not even grasp the metaphysical weight of the question.

Matthew Lee Anderson notes something similar when it comes to IVF: “One of my takeaways from the past decade of writing about IVF is that the sort of appeal Ross makes here sounds like a foreign language to most people—or maybe like that unintelligible teacher in Charlie Brown.”

I suspect my question about education sounds equally foreign. And so the only answer is to show by experience what virtue-driven education looks like, feels like, and produces. Taste and see that such education is good. That may be the only way forward.

But few have the courage to do so. Jennifer Frey attempted it at the University of Tulsa, with success. Yet administrators still shut the program down. Why? Possibly because the game is already lost. Education today is technological, technique-driven, and aimed at transferring data efficiently for economic ends. Go to school to get a job. That, it seems, is all that remains.

But might there be a more humane and enduring reason? I think so.