AI Ain’t Gonna Save Higher Ed
HOBOKEN, JUNE 14, 2026. It’s hard, no, impossible, to keep up with the flood of rants, I mean reflections, on how AI is corrupting, I mean affecting, science and education. So I’ll just mention a few things that have me brooding.
Last month “Ian,” a student in my first-year humanities course, gave me a paper headlined, “Will We Need Human Math Teachers?” Ian confesses he struggled in math courses last fall and this spring. He had a hard time understanding his professors and teaching assistants, or TAs.
Ian was too embarrassed to ask questions in class, and he didn’t want to slow the class down, so he turned to ChatGPT for help. ChatGPT answered Ian’s questions clearly, patiently, immediately, whenever Ian turned to it.
“AI gave me what no traditional classroom could,” Ian writes. “It gave me a learning environment that was judgment-free and paced exactly to my liking. But most importantly, I was able to ask as many questions as I wanted without slowing down other people or embarrassing myself.” Ian not only passed his math courses, he aced them.
Ian cites evidence to back up his claim that AI tutors work. He cites a 2025 study that separated Harvard students taking an introductory physics course into two groups: one received human tutoring, the other AI tutoring.
The AI tutor “helps students learn significantly more in less time and feel more engaged and motivated compared with in-class active learning,” the researchers report in Nature. “This study confirms the feasibility and effectiveness of AI tutors in educational settings.”
Here’s how suspicous I am: At the end of their paper, the authors acknowledge writing up their results with help from ChatGPT, which provided “surface-level grammatical input.” Can I trust researchers who rely on AI to evaluate it fairly?
Just as AI firms have formed “partnerships” with the media, so they are funneling money to influential scientists and mathematicians. Columbia mathematician Michael Harris says in Boston Review that the firm Handshake AI recently offered him $185–$400/hour “to help train an AI model to read, interpret, and summarize peer-reviewed scientific literature.”
Harris (whose Substack “Silicon Reckoner” I highly recommend) calls the Handshake gig “a textbook case of alienated labor,” Marx’s term for work that is degrading, meaningless and even harmful to the worker. AI firms are paying human experts to help the firms build AIs that replace human experts. The companies no doubt also expect the experts to serve as shills for AI.
Speaking of shills: Ian, my student, quotes mathematician Terence Tao extolling AI TAs. In a 2024 discussion Tao says: “You have 24-7 office hours where you can talk to a teaching assistant... if you're not supervised, you can get stuck at a step and there’s nobody to nudge you.” Tao predicts AI TAs “will become very widespread.”
Tao is also asserting that AI has become indispensable for mathematical research. This year, Tao told Nature: “A graduate student who refuses to touch AI systems and just wants to prove things the way we’ve done in the past might find they have fewer opportunities, unfortunately.”
Tao, to be fair, emphasizes that AI must be incorporated into mathematics cautiously. He has endorsed the so-called Leiden Declaration, in which an international group of mathematicians warns that “AI is putting fundamental values of the discipline under threat.” These values include transparency, openness and a dedication to comprehensibility.
“What happens if a mathematical proof is no longer the work of a human,” the Leiden group asks, “but of a proprietary AI model that academic researchers cannot access? Who is responsible for errors, and who gets credit if it is correct? And how can we tell whether an AI-generated proof is truly new, or simply a clever reformulation of existing work without proper attribution?”
The group recommends measures to ensure that AI does not harm mathematics or society in general. Noting the role of math-based technologies in “warfare, oppression, mass surveillance, and the undermining of democracy,” the Leiden group urges mathematicians to “evaluate the ethical consequences of your research” and “withdraw from harmful work.”
Imagine if all mathematicians, scientists and engineers, including those working on AI, adopted this ethical stance! But that is vanishingly unlikely. AI corporations are war profiteers--as you would know if you read the Pope’s recent critique of AI, which deplores the “unceasing development of weapons systems, particularly those involving AI.”
The Leiden group’s main concern seems to be that AI-enabled math prioritizes automation and objective results over the subjective human experiences of exploration, learning and understanding. See my exploration of this issue here.
Other fields share this concern. According to “The Last Astronomers.” a disturbing report in the June 4 Science, veteran astrophysicists fear that as young researchers become more dependent on AI, they will “lose or never build their own math, coding, and reasoning skills.” This trend “could lead to nothing less than the death of astrophysics as a human endeavor.”
Back to Ian’s essay, “Will We Need Human Math Teachers?” Good question, which applies to professors in general. If students prefer AI tutors to humans, universities might start replacing professors with AIs. After all, many universities, including my own, are already struggling with decreasing revenues.
One factor is a decline in birth rates in the U.S., which has reduced the pool of potential college applicants. Colleges are facing an “enrollment cliff,” as The New Yorker puts it, and many will probably not survive.
Meanwhile, the Trump administration has made it harder for non-Americans to attend U.S. schools and has slashed federal grants to universities. The White House just announced a new policy that would “give political appointees the final authority to deny federal funding for research deemed inconsistent with presidential priorities,” a New York Times oped says.
My university, Stevens Institute of Technology, is betting that AI will help it navigate this turbulent period. Stevens has created an AI major and a new school focused on AI. Ironically, tech companies, as they become more reliant on AI, are laying off workers at an accelerating rate, according to The New York Times, so it’s not clear how many jobs will be available for those AI majors.
A final story, related by my student Ian. This spring, he thought he nailed his final math exam, but the TA gave him only 89 points out of 100. ChatGPT told Ian that he deserved a higher grade. Ian presented his case to his professor, who said the TA had made a mistake and gave Ian a 96, enough for Ian to get an A.
Ian concludes his paper: “Not only did ChatGPT help me learn the material in order to pass my exam, but it also corrected human errors that my TA made. It made me wonder, ‘Will we need human math teachers?’”
Ouch. Pity that poor overworked TA, an aspiring mathematician, probably, competing with ChatGPT, a technology marketed by a ruthless, ultra-wealthy corporation. I know AI can do some things very well, better than humans. That is why I see it as catastrophic for science, math and the institutions that sustain them.
Further Reading:
An AI Critic Talks to a Tech School
How AI Moguls Are Like Mobsters