AI Splits the American University in Two

“Hysteria over artificial intelligence has divided Americans into two fearful camps: those afraid that the technology will advance too quickly, laying waste to human endeavor if not humanity itself, and those afraid that the failure to embrace AI quickly will doom us to irrelevance or Chinese domination.

 

In few places has this debate been as feverish as in the American university.

 

But AI's role in the university is a rare case in which both the restrainers (who argue that AI harms learning) and the cheerleaders (who argue that universities will fall behind without it) are right.

 

The holders of these outwardly contradictory perspectives are talking about two different things when they use the word "university."

 

The restrainers mean the liberal-arts college: the Anglo-American institution that imparts to students the accumulated knowledge of mankind, teaching them to think in the process.

 

The cheerleaders mean the research university: the Germanic institution that serves as a factory for new knowledge.

 

For decades, these two institutions have coexisted -- sometimes harmoniously, sometimes uneasily -- under the same roof.

 

The AI era may force a separation between undergraduate liberal-arts education and research-focused graduate education.

 

Early returns on the effect of large language models on student learning are concerning. The Massachusetts Institute of Technology's Ad Hoc Committee on AI found that "overreliance on chatbots can have a range of significant negative consequences -- diminishing critical thinking, weakening memory, eroding confidence, and undermining mastery." Large language models are mental crutches that allow us to find and reproduce knowledge without understanding it and to pass that knowledge off (perhaps even to ourselves) as our own. As knowledge and understanding diminish, so does the quality of thinking.

 

It is therefore no surprise that in a study of 25,000 Chinese students, AI use improved scores on homework while causing a significant drop-off in scores on exams when the students lacked technology access. Or that according to an MIT study, students who used AI to assist in writing essays could recall almost none of what they had written. Even so, students around the country are using the tools to help write essays, complete math and physics problem sets, and otherwise present the illusion of knowledge.

 

Yet AI is as crucial to the enterprise of research as it is corrosive to learning.

 

 By 2024, according to an Oxford University Press study, 76% of academic researchers were using AI tools to help collect or analyze data.

 

 AI is as prevalent in the career fields with which academic research intersects. Take medicine, where more than 80% of physicians now report to the American Medical Association that they regularly use AI in their work.

 

There seems ample justification for Dartmouth President Sian Beilock's recent claim that "if colleges and universities sit out this revolution, we abdicate a sacred commitment to prepare the next generation of leaders to navigate an ever-changing, always complicated, and frequently perilous world."

 

American universities are in a bind. They must produce students who can use AI as a tool to amplify their thinking while still cultivating in those students' minds the human inventiveness that AI can never replace.

 

They need to protect undergraduate instruction from the harmful effects of AI. In the classroom, technology should be roped off from the teaching and learning of nontechnological concepts. As Case Western English professor Walt Hunter has argued, schools should "bring back the blue-book exam," making student grades depend mainly on in-person exams and supervised essays. Longer-term assignments should be subjected to thesis-style "defenses" in class or faculty office hours to ensure student originality and deep knowledge of the work.

 

More supervised learning for undergraduates will require new faculty to specialize in teaching, meaning that more college professors in the future will be preceptors rather than research academics.

 

Once or twice in their undergraduate careers, students in the hard sciences and social sciences, as well as those in certain humanities areas, should breach the AI cordon sanitaire by taking a required course on the use of AI in their field. For students in AI-relevant majors, one or two dedicated courses should be sufficient before engaging further with the technology during graduate study. These courses won't be necessary for all humanities students. Familiarity with AI won't be any more necessary to write poetry in 2076 than it is in 2026.

 

Some worry that students will be unprepared for AI-powered research work in graduate school and the workforce unless their undergraduate years have been suffused by AI. There is little evidence to support this. If anything, the chief lesson of humanity's early encounters with AI is that it is easy to pick up and integrate the technology with no formal instruction in how to do so.

 

If the status quo continues unaltered and undergraduate instruction remains wedded to research, the risk to American intellectual leadership isn't that students won't know enough AI; it's that they'll know nothing but AI.

 

To borrow a phrase from Ms. Beilock, universities have a "sacred commitment" to ensure that after the AI revolution has passed, Americans can still think.

 

---

 

Mr. Masko is a freelance writer based in New Hampshire. He has worked as a case writer at Harvard Business School and on the faculty of Marywood University.” [1]

 

1. AI Splits the American University in Two. Masko, John.  Wall Street Journal, Eastern edition; New York, N.Y.. 22 Oct 2026: A15.