MIT's 55-Year-Old Great Books Program Finds a New Selling Point: AI

A first-year learning community at MIT founded in 1970 is being positioned, in the Institute's own telling, as newly relevant because of artificial intelligence. According to MIT News, senior lecturer Linda Rabieh says AI has made the question of what it means to be human unusually vivid for students, who are asking whether intelligence is only mathematical or only an aggregation of information. The program, Concourse, accepts 50 first-year students a year and pairs great-books seminars with math and science requirement courses taught by its own faculty.

For anyone building or buying AI systems, the story is less about a campus program than about a claim being made with increasing frequency across industry and academia: that the hard part of deploying powerful technology is judgment, not computation, and that judgment has to be taught somewhere.

What the program actually is

MIT describes Concourse as one of its first-year learning communities, founded in 1970 so humanities, engineering, and science faculty could collaborate on curriculum. Students take General Institute Requirement courses in math and science alongside humanities seminars, with readings that MIT lists as including Plato, Homer, Aristotle, Wordsworth, and Coleridge. A photo caption in the announcement shows students discussing Ovid's Metamorphoses.

The structure is small and physically concentrated. MIT says the program has a dedicated space in Building 16 with a classroom, seminar room, lounge, and kitchen, and that students receive advising and attend community events through the year. A weekly Friday advising seminar brings students together over lunch to discuss assigned texts or hold debates. Fall topics focus broadly on the human condition; spring sessions turn to practical political and moral questions, with MIT citing gene therapy, immigration, and gun control as examples. One caption describes a student arguing about whether social media threatens democracy. Several debates are run with the MIT Civil Discourse Project, a collaboration MIT dates to 2023.

Assistant Director Sasha Rickard, an alumna of the program, calls it a "mini-liberal arts college within MIT" and says many students cite it as a reason they chose MIT over other schools. Johnnie Jones VI, a senior in mathematics who serves as a teaching assistant and debate fellow, describes the program's main gift as learning to work through ideas independently. Those are testimonials, not measurements, and MIT provides no retention, admissions, or career-outcome data to support them.

The AI framing, and its limits

Rabieh's line — that AI is "good for business. Our business, of humanistic thinking" — is the most quotable part of the announcement, and it is worth reading carefully. She is not claiming that Concourse teaches AI, evaluates models, or produces policy research. The claim is narrower: that the arrival of capable systems has made abstract questions about intelligence feel concrete to eighteen-year-olds, which in turn makes the seminars land harder.

That is plausible and also unfalsifiable from the material provided. MIT does not say whether applications to Concourse have risen, whether AI-related topics have been added to the syllabus, or whether the debate calendar has shifted toward machine learning questions. The two debate topics named in the piece — social media and democracy, plus the spring slate of gene therapy, immigration, and gun control — predate the current AI cycle. Readers should treat the AI angle as context the Institute is drawing around an existing program rather than evidence of a curricular change.

Rabieh's substantive pedagogical point is sturdier and relevant well beyond campus. She argues that MIT students excel at quantitative reasoning but often avoid political and moral questions because they cannot see a rigorous way through them; Concourse, she says, teaches a non-quantitative method for thinking about such questions. Anyone who has watched an engineering team try to resolve a fairness, consent, or content-moderation dispute by searching for the right metric will recognize the failure mode. The industry's standard response has been to build evaluation harnesses and scorecards. Those are useful, but they relocate the judgment call into the choice of metric rather than eliminating it.

Leadership change and the institutional context

MIT says Ford Professor of Political Science Lily Tsai has been named faculty director, succeeding Anne McCants, the Ann F. Friedlaender Professor of History, who stepped down after 14 years. McCants notes that faculty are normally siloed from colleagues outside their fields and says sustained cross-disciplinary conversation is a distinguishing feature of the program.

Tsai takes over as MIT expands connections between the School of Humanities, Arts, and Social Sciences and other fields. MIT names three vehicles: the Compass Initiative, which Tsai co-leads with other SHASS faculty; the MIT Human Insight Collaborative; and the creation of shared faculty positions within the MIT Schwarzman College of Computing. Of these, the shared appointments are the most structurally interesting, because joint lines change hiring incentives and promotion paths rather than merely convening people. MIT does not disclose how many such positions exist, which departments they span, or how they are funded.

Tsai frames the program as preparing students to be "responsible citizens and leaders" at a moment when the world is trying to ground the use of powerful technologies in principled judgment. That is an aspiration, not a result. Fifty students per year is a small intervention at an institution of MIT's scale, and the announcement offers no mechanism by which seminar habits translate into better decisions inside companies years later.

Why this matters outside Cambridge

Hiring managers increasingly say they want engineers who can reason about second-order effects of deployed systems. Vendors increasingly market governance tooling that assumes someone, somewhere, can articulate what the organization values. Both depend on a supply of people trained to argue carefully about questions that do not resolve numerically. Programs like Concourse are one upstream source of that supply; corporate ethics training and model cards are not substitutes for it.

The open question is scale and evidence. Great-books seminars with 50 students, shared meals, and a dedicated lounge are expensive per head and hard to replicate. Until institutions publish something beyond testimonials — admissions demand, persistence, where graduates land, how they handle ambiguous technical decisions — the case for this model as a response to AI remains a reasonable intuition rather than a demonstrated one.

What to watch

• Whether Concourse's syllabus and debate topics formally add AI-specific questions under Tsai, or whether AI remains background context for existing readings.
• Details on the shared faculty positions between SHASS and the MIT Schwarzman College of Computing: how many, in which fields, and on what funding terms. None were disclosed.
• Any outcome data MIT releases on first-year learning communities — application volume, retention, or longitudinal tracking — that would move claims about the program from anecdote to evidence.
• Whether other technical universities stand up comparable small-cohort humanities communities, and whether they can run them at larger scale without losing the trust effects MIT attributes to shared classes and meals.
• Whether employers begin asking for this kind of training explicitly in AI governance, policy, and product roles, as opposed to treating it as a nice-to-have.