Your Cart is Empty
Tal Fortgang
‘17
Artificial intelligence is forcing universities to confront crucial questions about the nature and future of higher education. Among these: What, exactly, should a university require a student to do for himself or herself? Universities have embraced laptops, smartphones, online databases and every other technology that promised to make learning more efficient. Generative AI now offers efficiency on a different scale. It can draft an essay in seconds, summarize a book instantly, and produce an argument on almost any subject on demand. Universities may be tempted to treat these capabilities as inevitable improvements in education. Students will be able to consume vast amounts of information, more than ever before. But like AI writing itself, such consumption would be vast but shallow. The more interesting possibility, then, is that AI will force universities to rediscover the value of intellectual work that cannot be outsourced.
The University of Chicago has begun taking that possibility seriously. In July, its Law School announced an AI strategy built around the view that students should learn both with AI and without it, ostensibly because these are different kinds of learning. In required first-year courses, students will generally be prohibited from using laptops, tablets and phones in class, with limited exceptions. The school says the policy is intended to cultivate listening, reasoning, participation and the ability to make arguments in real time. Elsewhere, students will be taught to use AI deliberately and responsibly, including through required first-year instruction and more advanced experimentation.
Chicago has extended the experiment beyond the Law School. The Social Sciences Collegiate Division has announced that its Core courses will generally prohibit classroom technology and ban student and instructor use of artificial-intelligence tools beginning this fall. The experiment reflects the sensible proposition that a useful tool can become a poor substitute for the activity it was meant to assist. A calculator helps with arithmetic, a search engine helps locate information, and an AI model can help generate and revise prose. None of those capabilities eliminates the educational value of learning arithmetic, conducting research, or writing an argument—all of which involve working through multifaceted reasoning to an endpoint of clear expression.
Making this distinction will only become more important as AI improves. A university that defines education as the efficient transmission of information will have difficulty explaining why students need professors at all. Machines will be able to explain legal doctrines, summarize a novel, compare competing theories and produce a respectable first draft—all respectfully, patiently, and quickly. Universities would be left selling an expensive credential that obscures the fact that machines increasingly provide the same services for free.
The stronger case for higher education rests on harder-to-measure abilities. Students should leave college better able to concentrate, reason, write, speak, judge evidence and change their minds when the evidence demands it. Those abilities develop through practice. They require the student to sit with a difficult passage, formulate an argument, discover its weaknesses, and defend it before someone who knows enough to object. Struggling itself is educational – it’s how the deeper, harder-to-measure abilities are built – and ultimately geared towards the kind of education we talk about when we talk about education being valuable. Removing all of it would be like going to the gym to watch someone else lift weights. You may learn some things, but you will not build your own muscles.
Humanities departments have the most to lose but could seize the opportunity to make the most of these developments. For years, language departments and the humanities have struggled to explain their value amid declining enrollment and growing pressure to justify every course in terms of immediate career utility. AI exacerbates the problem. It can summarize Shakespeare, explain Kant, imitate the style of a historian and produce an essay on nearly any assigned topic better than most college students. If the humanities exist to produce competent prose or transmit information about canonical texts, their future looks bleak.
The good news is that they don’t. The humanities teach students how to pay attention to another mind. They teach them to recognize an argument buried in a difficult text, to notice an ambiguity that a summary omits, and to ask why a writer chose a particular word with all its connotations. Each is an exercise in understanding human beings as individuals and as a collective. Reading a great work slowly can change a student’s understanding of human motives, political institutions, and our moral responsibilities. These areas of substantive knowledge are crucial to master for young people to grow into fully flourishing adults. Humanities administrators and professors need to think about how the incentive structures they set up encourage and reward that profoundly humane enterprise.
AI also raises a challenge to the centrality of academic freedom that deserves more attention. Academic freedom depends on creating classrooms in which students can encounter ideas and test their intuitive responses against arguments they have not previously considered. That is obviously incompatible with outsourcing responses to an algorithm.
Universities have become increasingly concerned with ideological conformity. Much of that concern has focused on whether students and professors feel free to express unpopular views. The AI question introduces additional dimensions. Intellectual dependence can arise even in a classroom where every viewpoint is formally permitted. For one thing, AI large language models sound like the voice of God, but they are more akin to the synthetic voice of their programmers—who are hardly without their own prejudices. There is a real risk that students migrate towards the median view of Silicon Valley technologists, who have machines making their views sound especially reasonable and authoritative.
Moreover, students who habitually consult an algorithm before forming their own judgments may become less willing to endure uncertainty, confusion and disagreement. The result could be a generation of highly assisted thinkers who have had comparatively little practice thinking unaided.
Chicago’s approach is promising because it preserves opportunities for both technological fluency and intellectual independence. Students will encounter AI as a professional tool. They will also encounter classrooms in which the technology stays outside the door. That makes sense. A law student should learn to use AI to research and organize information, just as a lawyer will. He should also learn to reason through a difficult question because it’s possible that at some point no machine will be available, likely there will be times when the machine is wrong, and nearly certain there will be times when a question requires judgment that cannot be reduced to a prompt.
Higher education has spent decades equating innovation with adding technology, increasing convenience and removing obstacles. AI may force a reversal. Some obstacles deserve to remain. Students should sometimes have to write the paragraph themselves. They should have to remember an argument without consulting a screen and answer a professor’s question before having the opportunity to ask AI for help. These are small acts of intellectual self-reliance, but they are geared towards producing capable citizens. A university is one of the few institutions with the time and authority to cultivate them systematically.
A university confident in its educational mission should be willing to ask which technologies make students more capable and which merely make student work easier. The distinction will become more important as machines become more capable. The universities that make it well may discover that the future of education depends on preserving the inconveniences that technology is most often designed to replace. After all, those exercises train the muscles higher education was meant to build in the first place.
Tal Fortgang ’17 is a Legal Policy Fellow at the Manhattan Institute, a regular contributor to PFS and a contributing writer at The Dispatch.
Comments will be approved before showing up.
As a professor who is known to dissent from progressive ideologies that are dominant at universities such as Princeton, where I’ve taught for nearly 40 years, I’m frequently asked by students for advice about how to navigate a campus they worry will be hostile to them. Some are pro-Israel, or politically and socially conservative, or religiously observant.
They fear being treated unfairly by faculty members who are hostile to their beliefs. Many are afraid of being regarded as outcasts by their more liberal or progressive peers. They worry about being excluded from academic opportunities or not treated equitably in social or other extracurricular activities. To these students I say, with regret: You’re right to worry. I’ve seen these things happen.
In the battle between Harvard and the federal government, Princeton is siding with Harvard.
On July 22, the University signed an amicus curiae brief supporting Harvard’s suit against multiple executive branch agencies over research funding cuts in a federal appeals court. The brief was joined by 35 other research universities and colleges, including the rest of the Ivy League.
On July 31, federal judge Zahid N. Quraishi issued an opinion mostly denying a motion to dismiss a civil lawsuit brought by David Piegaro ’25. The motion was jointly filed by Princeton University and Kenneth Strother, the University’s police chief, who are the defendants in the case. The ruling sends 12 of the 13 claims against the defendants to discovery, where the parties will gather evidence before a potential trial.
The lawsuit stems from an altercation between Piegaro and Strother that occurred during the 2024 Gaza Solidarity encampment. Piegaro was charged with simple assault and trespassing after the incident. In April 2025, he was found not guilty of simple assault while the trespassing charge was dropped earlier that year in February. Quraishi’s opinion allowed most of the claims Piegaro made to proceed.
