Generative AI and Universities ―The Raison d’Être of a University

In November 2025, people's attention was focused on a post on the social media platform X. The post was from an account believed to belong to a university student^[i]. In essence, it read that "assigning tasks to generative AI produces better results than I can, and I feel powerless when overwhelmed by friends who can skillfully use AI." Such sentiments are likely common among university students. Generative AI is increasingly permeating even the research community. There is an anecdote of a graduate student with almost no background in economics presenting a professor of economics at the University of Tokyo Graduate School with an outstanding research paper which had been written with the help of generative AI^[ii]. AI learning support systems, such as OATutor developed by UC Berkeley, have become widespread, and a survey by the Digital Education Council^[iii] found that 86% of university students worldwide use generative AI on a daily basis. Amid these changes, people are beginning to seriously ponder if universities are still necessary and if AI might replace universities.

In the first place, what value has traditionally been provided by universities? In our discussion, let us classify that value into the four categories of (1) transmitting knowledge, (2) credentialing (conferring academic degrees), (3) conducting research, and (4) socializing and forming networks. Regarding (1) transmitting knowledge, much of this value is already becoming replaceable by AI. Indeed, AI can now replicate or even enhance most of the roles that have traditionally been filled by teachers. Some examples include tailoring instructional materials to individual students' proficiency, maintaining a systematic curriculum, assigning appropriate tasks with prompt grading, and providing immediate responses to questions. The value of (2) credentialing is also being called into question. In the United States, student debt at graduation has reached high levels. At the same time, notable cases have emerged among young people who, having dropped out of Stanford University, have amassed considerable wealth through AI development or fintech companies^[iv]. Such cases are not the exception. Major companies such as Google, IBM, EY, PwC, and Accenture have removed degree requirements for employment, instead placing emphasis on practical skill tests in AI and machine learning or on portfolios as hiring criteria. These trends are accelerating. With the advancement of AI, it is becoming increasingly difficult for university graduates in the United States to find employment. In fact, the unemployment rate among degree holders has risen to nearly 6%^[v] as of June 2025, reflecting a decline in the symbolic value of university degrees. The situation is similar with regard to the function of (3) conducting research. AI has become capable not only of reviewing literature, analyzing data, and assisting in writing papers, but also of generating research ideas. As illustrated by the anecdote from the University of Tokyo mentioned in the introduction of this paper, even individuals with limited specialized knowledge can use AI to produce results comparable to those of researchers. Of course, it would be premature to conclude from such cases that universities no longer provide value. Nonetheless, it is a fact that the core functions that universities have long performed are being partially supplanted by AI.

Even so, there is a perspective in which universities retain their unique value. In particular, the domain of (4) socializing and forming networks is said to be difficult to replicate with current AI, as it fundamentally involves embodied and serendipitous experiences^[vi]. At universities, countless experiences arise beyond academics: networks that lead to employment, long-lasting friendships, romantic relationships, and other significant encounters in life. Even if knowledge and research are surpassed by AI, universities will retain their value as spaces where human beings interact.

So, what does the future hold for universities under these circumstances? Looking ahead over the next decade, it is possible to envision three different scenarios.

Scenario 1: An era of massive university bankruptcies

If the value of transmitting knowledge and conferring degrees were to decline, the necessity of attending university would diminish. If harnessing generative AI makes it possible to quickly acquire skills equivalent to or exceeding those gained from four years of university study, many people may choose to enter the workforce without going to university. This would cause the demand for universities to drop sharply, leading to a future in which the majority of these institutions are eliminated.

Scenario 2: A select group of universities transform into ultra-elite institutions.

In an era in which knowledge and research functions are supplanted by AI and effectively available for free, the value of universities would shift toward the scarce experiences that are characterized by physical presence, serendipity, and human interaction. Universities that offer such experiences at high tuition fees would survive, potentially leading to a situation in which only the children of affluent families can attend. Effectively, universities could transform into membership-based clubs. In Japan, similar trends are already beginning to appear in some universities, as well as in private junior high and high schools.

Scenario 3: Close collaboration between corporations and universities

Collaboration between industry and universities could become a mainstream model in the future. However, this approach would carry its own risks. In faculties or research laboratories closely tied to specific companies, employment opportunities for students may be virtually limited to those firms, making it difficult to change to other industries. Dropouts would be common. Moreover, academic disciplines deemed unnecessary in the short term due to corporate demand may be reduced or even eliminated.

None of these three scenarios can be described as optimistic. One scenario envisions a future in which only the children of affluent families can "purchase" human relationships and experiences at universities, while students from the middle and lower strata belong to universities as quasi-employees who are constrained by corporate demands. Another scenario paints a future in which large numbers of universities disappear, and AI assumes full responsibility for individual education. Both images differ sharply from the traditional vision of the university as a place where diverse individuals gather to build relationships, grow through research and education, and ultimately graduate with degrees which are the keys to entering society.

The rapid development of generative AI is forcing us to reconsider the very raison d'être of universities. What functions should universities retain, what should they relinquish, and how should they transform? Some argue that confronting these difficult questions presents universities a chance for rebirth. Yet, at present, the future of universities remains shrouded in uncertainty. Perhaps, we should ask generative AI how to design the university of the future--that would be a paradox, wouldn't it?

---

[i] For the post on X, see the following URL.
https://togetter.com/li/2626646
[ii] Ogawa, H. "Tansei Review, No. 1184: A Graduate Student Arriving with AI in Hand," Campus Newsletter, No. 1599.
https://www.u-tokyo.ac.jp/gen03/kouhou/1599/end.html
[iii] See the following URL.
https://www.digitaleducationcouncil.com/post/what-students-want-key-results-from-dec-global-ai-student-survey-2024
[iv] See the following URL.
https://www.bloomberg.com/news/articles/2019-07-01/how-two-college-dropouts-built-an-860-million-fortune-by-age-23
[v] See the following URL.
https://www.bloomberg.com/news/articles/2025-06-09/college-grads-face-tough-job-market-as-trade-war-ai-create-uncertainty
[vi] That said, it is highly likely that such value will also be offered in settings outside the university. Moreover, whether AI is inherently ill-suited to socialization and network-building remains a matter of debate at present.