What Comes After the Multiversity?

An MVP for the post-multiversity college

In my piece yesterday for Persuasion, I argued that the American “multiversity” is finished. The great mid-century high modernist synthesis of research, teaching, credentialing, and coming-of-age all bundled under one institutional roof has been fraying for sixty years, and the arrival of LLMs heralds its death knell. When a student can produce a plausible term paper in twenty minutes using Claude or Gemini, when an AI tutor can explain any concept at any level of sophistication with infinite patience, when AI can ace most professional examinations, the inherited architecture of the university has stopped working. The lecture, the term paper, the credential: these load-bearing technologies of the postwar university are all collapsing at once.

AI is exposing tensions that were always present in the multiversity: between research and teaching, between credentialing and learning, between the lecture hall and the fraternity party. But I argued in Persuasion, this doesn’t have to be a catastrophe.

The opportunity is disaggregation: let the functions go their separate ways, and recover something closer to the classical model of education in the parts that remain. Developed in the tutorial and the seminar, with faculty as interlocutors in the seminar room, the new pedagogy should focus on helping students to develop cognitive capacities like epistemic self-regulation, goal constitution, and the cultivation of taste. These are exactly the things AI can’t do, and they’re exactly what serious education was always supposed to develop. And the curriculum to build them, as I argued earlier in Noema, is the classical liberal arts, that is, the combination of literature, philosophy, and history; psychology, sociology, geography, and political science; biology, chemistry, physics, and earth sciences; and mathematics and computer science.

The central problem is that moving to this model is going to be fiendishly difficult for existing universities, who face a classic case of the innovator’s dilemma. As I say in the Persuasion piece:

Elite universities in particular have built their faculties almost entirely around research achievement, with teaching treated as a secondary obligation. Reconceiving the professoriate will mean altering tenure criteria and promotion incentives, and it will face fierce resistance from scholars whose professional identities are bound up in the research function. No doubt some tenured faculty will pour boulders and boiling oil down the side of their ivory towers to prevent these changes from taking place.

A minimum viable product for higher ed in the age of AI

What I didn’t say in the piece is that instead of reforming existing universities, it may turn out to be easier to build new higher education institutions that are optimized from the ground up for the age of AI. So let me try to sketch what I think an institution built on these principles might actually look like.

First of all, recognize that there are many features of the multiversity that will never be part of the new institutions: sports teams, dead-tree libraries, museums, faculty-oriented laboratories, fraternities & sororities, co-curricular bureaucracy, etc. All this stuff, which dramatically increase both the costs and operational complexities of universities, will be designed out the new institutions from the jump. (This is what I mean when I use the term “minimum viable product.”)

Second, start small. A founding cohort of seventy or eighty students. A two-year program, since an MVP doesn’t have to do everything before it’s proved it can do anything. No need for accreditation, which is a self-referential legacy of credential-mindedness. One physical site, with all students and a core of resident tutors living on it. Communal meals. Weekly town meetings for governance. The model is the Platonic academy crossed with the Oxford tutorial system, with a working kitchen at the center of the building.

Third, hire faculty who above all are excellent interlocutors. A dozen or so faculty, paid well (say $150K + rooms/meals), not expected to publish, evaluated on student outcomes by outside oral examiners. A mix of academics, practitioners, and writers — all of whom are themselves AI-savvy. The job is to sit with students in pairs and small groups, to read the same texts students are reading and dialog with them in person, to model how to calibrate uncertainty in real time, and to call out sophistry and bad reasoning when it shows up. For the faculty, the professional incentive is the pedagogic work itself and the company of serious colleagues. It should be a Beruf, in Max Weber’s sense.¹

The curriculum should consist of four classical strands: humanities, social science, natural science, formal science. A shared reading list moves each cohort through together (yes, a canon!), with concentration choices opening up in the second year. Threaded through everything, a fifth strand: working with AI. Instead of treating LLMs as a problem to be policed, treat them as tools, objects of critical study, and collaborators students will rely on for the rest of their working lives. Prompt construction becomes an exercise in goal constitution. The model’s explanation of Hegel gets workshopped against the text itself. Students learn what the machine does well, to remain vigilant for error, and how to recognize the difference under pressure. Think Deep Springs, optimized for the age of AI.

Four assessment instruments only. Group projects with decision logs and prompt logs submitted as supporting documentation. Handwritten exams in a proctored room, twice per term. Oral defenses before two-tutor panels, monthly. And a capstone project defended before outside examiners and the rest of the student body. No term papers. No grades on the transcript. Graduates leave with a portfolio: their exam scripts, video of the capstone defense, narrative evaluations from each tutor.

The bet is that employers in AI-forward sectors will value verifiable cognitive demonstration over a credential whose signal value is eroding anyway. A smart institution would get these sorts of corporate partners involved from the get-go, partly as financial supporters, but more importantly as job placement sites. If you know that attending this newfangled institutions means that hiring managers at McKinsey, Goldman, Google, and SpaceX are at least going to give your resume a look, that alone will draw in a host of talented students.

Who would go to a place like this?

Not most students, at first. Most conventionally ambitious students are likely to still seek the signaling value of a traditional elite university degree. The bill of goods sold by contemporary American higher education in every way is designed to perpetuate the idea that a traditional four year college degree is the only viable path up. Because the most talented and ambitious students continue to buy it, has become a self-fulfilling idea — one reinforced by the hiring practices of top organizations, which in practice focus heavily on recruiting from traditional high-prestige colleges. That model is wobbling, but it still works for many people, and a small new institution doesn’t have to compete for them.

Despite the stolidity of that cultural believe about universities, there is a real and growing constituency this new sort of institution can speak to. There are students who already understand that the value of a credential as such is hollowing out — students who maybe watched their older siblings struggle to find purchase after expensive Ivy degrees, who use AI all day and have seen up close how it changes what’s worth knowing, and understand that knowing how to run a team of agentic AI models (and not be bamboozled by them) is the job description of the future.

There are also families who care about formation more than credentialed status, who want their kids to come out of college above all knowing how to think and argue and read difficult books. There are students from outside the traditional feeder schools who’ve been told for years that the meritocratic ladder runs through the same fifty institutions, and who are starting to suspect this is no longer true. There are the brilliant autodidacts who would otherwise skip college entirely, and who’d be drawn to a place that takes their seriousness seriously.

The students who’d be wrong for this model are also easy to identify. Anyone whose primary goal is the credential as such. Anyone looking for a low-friction four-year coming-of-age experience subsidized by parental anxiety. Anyone whose intellectual self-image depends on being lectured at by famous people.

The students who’d be right for it are the ones who’d already half-figured out, before reading any of this, that the inherited model isn’t going to deliver what they’re really after. There aren’t that many of them yet. But there’s going to be many, many more going forward. Serving them is an enormous entrepreneurial opportunity.

1. Beruf in German means both “profession” and “calling,” and the tension between these two dimensions of Beruf is central to Weber’s analysis of the tragedy of modernization — a tragedy that was decidedly read out of American liberal-technocratic social scientific view of Weber that the postwar multiversity both championed and embodied.