1. The Lecture Is Losing Its Monopoly
For a very long time, one of the simplest reasons to go to college was that the people at college knew things that you did not. If you wanted to learn something difficult, you went somewhere that had knowledgeable teachers, books, laboratories and a structured curriculum. The professor was not merely a guide. In many cases, the professor was the scarce source of knowledge. A student who wanted to understand quantum mechanics, economics, computer science or molecular biology needed access to someone who had already spent years mastering it.
The internet weakened that monopoly, but AI weakens it much further. Search engines could find information. AI can increasingly explain it, reorganize it, personalize it, quiz you on it and continue teaching until you understand it. You are no longer simply downloading information from the internet. You can have something resembling an always-available tutor that adapts to your level and responds to your questions.
AI can also compress the distance between disciplines. A student studying biology can ask for programming help, statistical analysis, a literature explanation and a simplified explanation of a mathematical concept without having to find four different specialists.
This does not make professors irrelevant. Good professors do much more than transfer information. But it weakens the old economic justification for the classroom: the professor is no longer the only intelligent entity available to the student.
This is not merely a thought experiment happening outside universities. MIT is openly wrestling with exactly this problem. In August 2026, MIT President Sally Kornbluth described AI and education as a “watershed moment” for MIT and higher education. MIT’s own review of AI in teaching, learning and research called for changes to assessment, greater emphasis on hands-on learning and AI policies suited to individual courses.
MIT: AI and education — A watershed moment
That is a revealing response. MIT is not simply asking how to teach students to use AI. It is asking what needs to change about education itself.
For students who are already highly motivated, the implications are even larger. A curious student with a laptop and a capable AI system can increasingly construct a personalized education outside the university. That leads to the central question of this article: if AI can provide much of the intellectual assistance that once required a college, what exactly is the college now selling?
2. What Are You Actually Buying?
Once knowledge becomes abundant, we should start looking at what remains scarce rather than what has become cheap.
A student is not buying only lectures. They may be buying access to professors, exceptional peers, laboratories, research equipment, institutional reputation, internships, alumni networks and opportunities that are difficult to reproduce independently.
The distinction between knowledge and environment becomes important. Knowledge may increasingly be available everywhere, while environments remain local and expensive to build. An individual can ask AI to explain physics. That does not give them access to a particle accelerator. AI can explain how scientific research works. That does not automatically put the student inside a serious research group. AI can help write code. That does not create a team of talented people building something together. AI can simulate an interview. It does not give the student a relationship with an actual employer.
A degree also performs a signalling function. Employers do not merely want to know whether you know something; they want evidence that you have demonstrated persistence, competence and the ability to operate within a demanding system. Whether the degree remains the best signal is another question.
There is an interesting MIT study that adds another dimension to this discussion. Economist David Autor and his colleagues examined how new forms of work emerged in the postwar American economy and found that new technology-driven work historically benefited young and educated workers disproportionately. But they also found that the scarcity value of new expertise tends to erode as that expertise becomes common or gets automated.
MIT research: Technology usually creates jobs for young, skilled workers. Will AI do the same?
That is important for college because it suggests that the value of education has never been completely static. Universities have historically trained people for emerging forms of work. The problem is that AI may accelerate the cycle: today’s scarce expertise can become tomorrow’s ordinary capability much faster.
The question is therefore not simply whether college remains useful. It is which parts of college remain scarce enough to justify the cost. This creates a useful distinction between what college teaches you and what being at college gives you. The first category is increasingly vulnerable to AI and the internet. The second may remain valuable precisely because it depends on scarcity.
The uncomfortable possibility is that the value proposition of college could gradually shift from “we have knowledge you cannot easily obtain” to “we have people and opportunities you cannot easily access.”
3. College as a Human Clubhouse
This brings us to a part of college that economists and education brochures tend to describe rather badly: college is also a concentrated social environment.
Young people spend several years surrounded by other young people who are simultaneously figuring out careers, relationships, identity, interests and ambitions.
Friendships are an obvious part of this, but the network is much broader. Classmates become colleagues, collaborators, founders, researchers, competitors, spouses, investors and professional contacts.
Professors matter here too, not simply as sources of information but as people who can notice talent, introduce students to opportunities, recommend them for work and sometimes change their trajectory.
Clubs, sports, cultural activities, college festivals, informal conversations, shared projects and even arguments in cafeterias create experiences that are difficult to reproduce through an AI interface.
Dating belongs in this discussion too. College has always been one of the places where people meet potential partners, even though universities rarely advertise this as part of their value proposition.
None of this means every campus automatically provides an extraordinary community. A mediocre social environment is not magically valuable because it has a campus. The quality of the people and the density of meaningful interactions matter.
But if AI increasingly handles the individual intellectual side of education, these human interactions may become relatively more important.
Interestingly, MIT has arrived at a similar institutional concern. Its 2026 AI review explicitly calls for MIT to center people, community and the residential experience in response to AI, alongside changes to teaching and assessment.
That is a significant clue about where universities may be heading. As machines become better at delivering cognitive assistance, the physical and social environment may become a larger part of what the institution is actually providing.
An Indian professor is asking a very similar question. Prof. Prathosh A.P. of IISc recently had an informal conversation with his students titled “Pedagogy in the Times of AI,” explicitly asking what the purpose of higher education should be in the age of AI and what needs to change in universities and pedagogy.
Prof. Prathosh A.P.: “Lec 16 — Pedagogy in the Times of AI”
His later reflection on teaching in the AI era makes the shift even more explicit. He has argued that when he began teaching, the challenge was helping students access knowledge; today, knowledge is abundant, and the bottleneck is developing judgment. He sees the teacher’s role moving toward helping students ask better questions, reason from evidence, design experiments and critically evaluate AI-generated outputs.
The idea of the human clubhouse therefore becomes useful. The campus may increasingly function as a place where people meet, collaborate, compete, experiment, socialize and form relationships while AI handles a growing share of the cognitive work.
4. But Who Gets the Clubhouse?
We should not romanticize this transformation. If the scarce resource becomes access to exceptional people and environments, scarcity may become even more important.
A student can potentially get world-class explanations from AI without attending an elite university. But they cannot necessarily get the same access to a particular professor, laboratory, research group, alumni network or social circle.
This creates an interesting possibility: AI could democratize intellectual capability while elite education continues to concentrate social and institutional capital.
The richest students may therefore have both: powerful AI available to everyone and privileged access to the physical environments that AI cannot reproduce.
This could make the traditional distinction between elite and ordinary colleges more complicated. If everyone can access roughly similar AI tutors, the differentiating factor becomes the quality of the human environment surrounding those tools.
A university might therefore become increasingly valuable because of its people, reputation and access rather than because its professors possess information unavailable elsewhere.
That raises an uncomfortable question about the price of higher education. If a large portion of the educational content can be obtained independently, how much should students pay for the remaining bundle?
It also raises a broader question about social mobility. If education shifts from distributing knowledge to distributing access, does higher education become more egalitarian—or simply a more sophisticated mechanism for selling access to privileged networks and environments?
There is a strange irony here. AI may make the intellectual part of education more equal while making the social part more valuable and therefore potentially more exclusive.
5. The Student Changes
The student in an AI-enabled university is fundamentally different from the student for whom the traditional curriculum was designed.
Instead of waiting for a professor to explain something, the student can explore independently and arrive in class with much more sophisticated questions.
AI becomes tutor, research assistant, coding partner, editor, simulator, language teacher and brainstorming partner.
This can dramatically increase the speed at which an ambitious student moves through basic material. A student no longer has to wait until next week’s lecture to resolve a question.
But greater capability does not automatically produce greater achievement. Once everyone has access to powerful assistance, the advantage shifts toward the ability to use it intelligently.
Students have to learn how to distinguish a useful answer from a plausible-looking one, formulate good questions, verify claims, compare alternatives and decide what deserves attention.
There is also a new problem of abundance. Previously, students often struggled because they could not find enough information. Now they can drown in information, explanations, possible projects and generated ideas.
AI therefore attacks information scarcity while potentially creating information fatigue. The bottleneck moves toward judgment, selection, curiosity and purpose.
This is remarkably close to the argument Prof. Prathosh makes in his own reflection on teaching in the AI era. His point is not that teachers have become unnecessary because AI knows more. It is almost the opposite: when information becomes abundant, helping students develop judgment becomes more important.
The student who benefits most may therefore not be the person who knows the largest number of facts. It may be the person who can combine AI’s enormous cognitive capacity with a clear sense of what is worth doing.
This changes the meaning of education itself. Education becomes less about accumulating answers and more about developing the ability to choose questions, evaluate answers and act on the results.
6. Research Changes Too
The transformation does not stop at teaching. Universities have traditionally justified themselves partly by being places where new knowledge is created.
AI is now entering that process directly.
Researchers can use AI to search literature, compare papers, identify connections, generate hypotheses, write and analyse code, process data and explore possible explanations.
This is not a distant possibility at MIT. In her August 2026 statement, Sally Kornbluth noted that AI tools were already dramatically expanding the scope and pace of experimental research at MIT, including helping researchers generate promising hypotheses and pressure-test possible solutions.
That makes the research question much more interesting. AI is not merely something students use to write papers. It is becoming part of the process through which universities themselves create new knowledge.
As these systems improve, the speed and scale of research can change significantly.
This creates the same structural problem seen in teaching. If AI can perform more of the intellectual labour involved in research, the scarce contribution shifts upward.
The important researcher may increasingly be the person who identifies an interesting problem, designs a meaningful experiment, understands the limitations of the evidence and decides whether a result actually matters.
There is also a more radical possibility: AI could generate research directions that humans would not have considered themselves.
That changes the university’s role from being primarily a place where humans execute research to a place where humans and machines increasingly select and evaluate research possibilities.
Students could therefore be exposed to research much earlier because AI lowers some of the barriers to entering technical work.
At the same time, physical experiments, specialized equipment, serious research groups and expert supervision become more valuable because they remain difficult to reproduce outside institutions.
The university may consequently lose some of its monopoly over intellectual labour while strengthening its importance as a physical research environment.
7. The Great College Reductio ad absurdum
Now deliberately push the argument too far and see where it leads.
Imagine AI becoming capable enough to teach most standard academic material, help students practise it and assist with research and projects.
The university then faces a verification problem. If AI can write the essay, solve the problem and produce the code, how does the institution know what the student actually understands?
Assessment begins moving away from work that can simply be outsourced to AI and toward oral examinations, demonstrations, experiments, presentations, project defence and real-world application.
This is already part of the direction MIT is considering. Its AI committee has explicitly called for a reevaluation of assessment and greater emphasis on hands-on learning.
The university becomes less concerned with whether a student can produce an answer and more concerned with whether the student understands, can defend and can use the answer.
At the same time, the campus retains the things that remain difficult to digitize: laboratories, professors, exceptional peers, relationships, clubs, social life, networking and reputation.
Push the logic far enough and college begins to look strangely different from the institution we know today. It might increasingly resemble a combination of assessment centre, research laboratory, networking environment and human clubhouse. And, because we are being deliberately provocative, we can add dating to the list.
The point is not that colleges will literally become dating clubs. The point is that once AI removes some of the traditional educational functions, the supposedly peripheral functions of college become much easier to see.
The joke therefore exposes a serious issue: if the lecture is no longer the main product, what is the product?
8. So Why Go to College?
Return from the reductio and answer the question more seriously.
There are still many good reasons to attend college: access to difficult physical environments, exceptional peers, professors, research opportunities, professional networks, credentials and concentrated social experience.
But these reasons vary enormously from one institution to another.
The future student therefore cannot simply ask which college has the best curriculum. They have to ask what part of the college experience is genuinely difficult to reproduce independently.
For some students, the answer may be a laboratory or research programme. For others, it may be an unusually strong peer group or a professional network. For others, it may be the credibility of the institution. For some, the social environment itself may be an important part of the experience.
And for some students, perhaps the honest answer will be that a particular college does not provide enough additional value to justify its cost.
That would be a significant change. The question would shift from “Which college should I attend?” to “What can this college give me that I cannot efficiently obtain with AI, the internet and my own initiative?” Universities will increasingly have to answer that question too.
9. The Bigger Question
The discussion can now broaden beyond college and connect the issue to the larger transformation of society.
College is one link in an old chain: education leads to qualifications, qualifications lead to jobs, jobs lead to income, income leads to status and financial security.
That system worked partly because intellectual skills were scarce and employers needed institutions to identify people who possessed them.
AI potentially disrupts several links simultaneously. If AI increases the productivity of a small number of highly capable people, companies may need fewer conventional knowledge workers. If AI makes many forms of expertise widely available, the value of merely possessing knowledge may decline. If credentials become less reliable indicators of capability, employers may increasingly look at demonstrated results. If work itself changes, the economic reason for spending four years preparing for a specific occupation becomes less certain.
MIT’s David Autor research is useful here because it provides an important counterpoint to the simplistic “AI will destroy all jobs” story. Historically, technology has not only eliminated tasks; it has also created new forms of work, and those new forms of work have often disproportionately benefited young, educated workers. But Autor also points out that the scarcity value of new expertise eventually erodes as that expertise becomes common or automated.
That makes the future of college less obvious than either the optimists or the doomers suggest. College may still be a powerful route into new forms of work. But it cannot assume that the expertise students acquire at eighteen will retain the same scarcity value at twenty-two.
And if education becomes less about preparing people for a stable job and more about helping them navigate an unstable technological environment, the purpose of college becomes much broader.
This is why the question is not really “Will AI kill college?” The deeper question is whether the old education → employment → income → status bargain remains intact.
College may survive perfectly well while the reason people go to college changes dramatically.
10. The College We Inherit vs. the College We Need
The inherited college is largely organized around scarce information, scheduled teaching, standardized curricula, conventional examinations and credentials.
The emerging college has to operate in a world where information and cognitive assistance are abundant, curricula can change continuously and students have powerful AI systems beside them.
Its value will therefore have to come increasingly from things that abundance does not automatically provide: exceptional people, difficult environments, physical research, trust, reputation, relationships, collaboration, judgment and meaningful experiences.
This does not mean abandoning knowledge. Fundamentals may become even more important because students who understand the underlying structure of a field can use AI much more effectively than those who simply accept its outputs.
MIT’s current approach points in precisely this direction. Its response is not to abandon education or retreat from AI. It is to combine AI-aware education with hands-on learning, human community, residential experience, continuous adaptation and stronger attention to how learning is assessed.
The university’s role may therefore move from distributing intelligence to helping people use abundant intelligence well.
That is a much more demanding mission than delivering lectures.
And it brings us back to the original question. If AI gives almost everyone a pipeline to the sea of knowledge, what exactly makes a particular university’s shore worth visiting?