AI & Higher Education
AI didn't break the university. It found it already broken.
Universities spent forty years justifying themselves in the language of job preparation and economic productivity. AI is now calling that bluff. The right response is not panic or capitulation — it is clarity about what education was always actually for, and what AI can and cannot do.
The institutional argument — that universities exist to produce employable graduates — generated two related but distinct consequences. The first — particularly ascendant at my former institution — was the idea that courses should be designed to produce outputs that might be enticing to future employers. But this is only the curricular translation of a deeper problem: the idea that education was about forming students with dispositions geared toward delivering products per specifications on deadline. If you tell students that education is about producing things, you should not be surprised when they use a tool that produces those things without the intervening inconveniences and difficulties of thought. You should also not be surprised that they don't understand the joy of discovering a new connection, the pride of finding evidence that bears out their intuitions, or the deep pleasure of changing their own minds.
Students who submit AI-generated work are not "cheating" so much as complying with a system that has lost sight of its own purpose.
The argument
A bad-faith bargain more than forty years in the making
The purpose of college education was never job preparation
Education is learning how to ask new questions. The purpose of education — at its best, and in the tradition that produced the modern university — was human formation: the production of a different kind of person. Someone with the knowledge, the critical capacity, and the intellectual resources to live a richer, more examined life.
Universities made a bad-faith argument and gaslit themselves into believing it
From the 1980s onward, universities began justifying themselves in the language of human capital, workforce development, and return on investment — because that was the language that moved legislators, reassured anxious parents, and showed up in rankings. The people running these institutions mostly knew better. They made the vocational argument anyway. Having made it for forty years, many of them came to believe it themselves.
AI is not creating a crisis — it's revealing the misguidedness of the vocational argument
If the purpose of a degree is job preparation, and AI is reorganizing the labor market faster than any curriculum can track, the economic case for a university degree becomes genuinely difficult to sustain. Even now, AI can do passably well some of the tasks associated with liberal education — digesting information, for instance, or analyzing data. But this is only a crisis if you accepted the vocational framing. AI did not create this situation. It revealed it.
This is the moment to say what education is actually for
Because the vocational argument is now untenable, the ground is unusually open for a different argument — one that was always true but suppressed for decades by its political inconvenience. If you understand the purpose of education as human formation rather than credential production, AI doesn't threaten education at all. It might even, used well, serve it. But using it well requires being clear about what it is — and what it isn't. Education is learning how to ask new questions. AI by itself cannot teach this.
Getting the category right
What LLMs can and cannot do for education
Facilitate the work of thinking
LLMs facilitate the reproduction and recombination of existing meaning at dramatically reduced cost. Used well — as a partner in the process of thought rather than a substitute for it — AI can genuinely serve the formative purpose of education.
- Reviewing existing literature and recommending relevant contributions
- Analyzing very large data sets quickly
- Orienting you in a field not your own
- Summarizing and articulating positions and arguments
- Generating associative links across bodies of knowledge
- Stress-testing arguments and ideas
- Proposing clearer articulations of thoughts
Replace intellectual formation
An LLM can sometimes produce an adequate essay without a human having thought anything through. But of course the product was never the point; the process of producing it was.
- Asking importantly new questions or proposing new frameworks
- Discerning the relevance of an idea to a particular situation
- Vetting sources reliably
- Bringing human experience to bear on a particular issue or problem
- Critically assess truth claims, regardless of their source
- Understanding human flourishing as something other than consumption
Students who submit AI-generated work are not "cheating" so much as complying with a system that has lost sight of its own purpose.
— from "The Deliverables Problem"The political stakes
The vocabulary of rupture on campus
In higher education, the declaration that AI represents a new era of human history does work beyond what I am calling the retroactive foreclosure of political accountability. The university is an institution whose obsolescence is useful to those making the declaration — and the reasons are not hard to identify.
The university trained the researchers who built the models. It produced, funded, and published through its presses much of the writing that constitutes the training data. It credentialed the people whose expertise the AI industry trades on when it presents its outputs as knowledge. AI discourse declares the university obsolete using not just the knowledge but also the authority it extracted from the university in the first place. It is a structural incoherence that the rupture narrative depends on no one noticing.
Such a declaration obviously serves those who stand to profit from the delegitimization of these institutions. Those who stand to profit from replacing the university's functions — with AI tutors, credentialing alternatives, skills-based hiring, automated content — have every interest in deploying the language of rupture. The rupture narrative is self-sealing: the frameworks best equipped to contest it are the ones it has already subjected to retroactive foreclosure. The rupture narrative has already falsely mooted the terms on which that argument could be made.
What the university actually has — and what AI cannot replicate — is the capacity to produce people who ask questions that haven't been asked before; interrogate structures and institutions rather than simply operate within them; critically assess truth claims, regardless of their source; bring experience itself to bear on questions of value and meaning; and understand human flourishing as something other than consumption.
The affirmative case
The case for liberal education is stronger than it has ever been
The real problem is not that AI will make education impossible. It is that so many have lost sight of what education actually does.
If you understand education as human formation — the cultivation of people who can think carefully, interrogate structures and institutions, and bring their own experience to bear on questions of value — AI does not threaten it. Used well, AI can serve it: as a tool for research synthesis, for stress-testing arguments, for providing the kind of articulate resistance to half-formed ideas that makes thinking sharper.
What AI cannot do is produce in students the joy of discovering a new connection, the pride of finding evidence that bears out their intuitions, or the satisfaction of having worked through a problem rather than routed around it. These are not incidental features of education that might eventually be automated. They are what education produces — and what a population without them looks like is a political question as much as a pedagogical one.
The deeper irony is this: if AI delivers even a fraction of the productivity gains its advocates promise — if it frees human beings from the necessity of performing tasks that machines can perform — then the question of what we do with that freedom becomes the central question of our time. What is a good life? What is worth doing? How do we talk thoughtfully with each other about things that matter? These are not new questions. They are the oldest questions. On its own terms, AI makes liberal education not less necessary but more urgently needed than it has ever been.
Working together
Available for writing, consulting, and speaking
What I can offer
I spent thirteen years building a curriculum inside an urban institution, watching the liberal arts get systematically devalued — and then used as a budget line. That gives me an unsentimental, ground-level view of what AI is actually doing to universities, and what administrators, faculty, and trustees are getting right and wrong about it.
I'm available to write on these questions for general and specialist audiences, to consult with institutions thinking through their AI response, and to speak to faculty, administrative, and board audiences. For speaking engagements, see the Speaking page. I'm particularly interested in conversations with foundations and think tanks working on higher education policy.