Professors fear AI is making students forget how to think

At first, the problem looked familiar: Students were submitting work that was not really their own. Universities responded with syllabus updates, detection tools, disclosure rules and new assessment formats.

But the issue has quickly outgrown cheating. The harder question is what colleges still expect students to practice personally when software can produce fluent answers in seconds.

In The New York Times, Anastasia Berg, a philosophy professor at the University of California, Irvine, argues that even basic AI use can harm students because writing is not just a way to display knowledge.

In The Point, Megan Fritts, a philosophy professor at the University of Arkansas at Little Rock, approaches the crisis from inside university AI committees, where the debate is less about one violation than about what higher education is trying to protect.

The classroom problem is concrete

Berg writes that many students in a large general education course appeared to have used AI against her policy while completing final take-home exams. The error became visible because the assignment referred to recent philosophical work, while some AI-assisted answers drifted toward unrelated medieval theology with similar titles.

That example shows both the weakness and attraction of the tool. AI can make student work look complete while separating it from the course, the assigned material and the student’s own understanding.

At Fritts’s university, committees have discussed detectors, appeals, syllabus language, Grammarly, equity, environmental concerns and assignment design.

Her point is that these debates often circle around procedure. They ask how AI should be managed, but not always what losses universities are willing to accept.

Writing is more than a deliverable

The strongest warning from both professors is that universities may mistake the product for the education.

If the goal of a writing assignment is only a polished essay, AI looks like an efficient shortcut. It can draft a paragraph, organize an answer, clean up awkward wording and imitate confidence. But Berg argues that the value lies in the student’s own effort: Confusion, revision, failed phrasing and gradual clarification.

A rough paragraph can reveal a student trying to understand something. A smoother AI-generated paragraph may hide the absence of that struggle.

Fritts makes a related but more institutional argument. She suggests that humanities departments cannot defend themselves merely by saying they produce essays, arguments, poems or presentations.

If those artifacts are the whole point, AI will appear to threaten or replace them. If the purpose is the formation of people, the process cannot be outsourced without changing the course itself.

Universities are trying to catch up

The wider university response has been uneven because AI changes faster than academic governance. Many institutions have created committees, task forces and guidance documents.

Some encourage AI reflection logs or require students to explain how a tool was used. Others push oral exams, in-class writing, staged drafts or discussion-based assessment.

These responses are practical, but they also reveal uncertainty. A disclosure policy assumes AI use can be permitted if documented. An in-class writing requirement assumes some tasks must remain unaided. A detector-based system assumes the main problem is enforcement.

Fritts’s essay shows how difficult consensus has become even among people who care deeply about teaching.

Some instructors want students prepared for workplaces where AI will be normal. Others believe the classroom should be one of the few places where students practice without automated help.

The humanities face a sharper test

The debate is especially intense for the humanities because these fields are already under pressure to justify their value. Literature, philosophy, history and related disciplines are often defended in practical terms: they teach communication, interpretation, analysis and argument.

AI complicates that defense by producing language that resembles those outcomes. It can summarize a poem, outline a philosophical position, draft a historical comparison or generate a discussion post. To administrators focused on efficiency, that may look like a reason to integrate the technology more deeply.

Fritts warns that this product-based view leaves the humanities exposed. If the purpose is only to produce acceptable written artifacts, faster artificial production will always be tempting.

But if the purpose is to help students become people capable of expression and judgment, the artifact alone is not enough.

Some spaces may need stricter rules

A serious university response does not require pretending AI will disappear. Students will use it outside class, and many workplaces will expect familiarity with it. There are also legitimate questions about access, disability support, language barriers and professional preparation.

But accepting that AI exists is different from allowing it everywhere. Education has always depended on protected forms of practice. Students solve problems before seeing the answer. They discuss books before reading summaries. They draft, revise and defend claims before being told what an expert thinks.

For that reason, some courses may need firm restrictions, especially when the purpose is to develop a student’s own command of language. A humanities classroom can teach about AI, analyze AI and criticize AI without allowing it to replace the student’s first encounter with a text or idea.

The practical challenge is honesty. Universities should not praise critical thinking while permitting students to bypass the exercises that build it.

The question is not whether AI can write. It can. The question is whether students can still learn when writing becomes optional.

Sources: The New York Times, The Point

This article is made and published by Kenneth Glad, who may have used AI in the preparation