For months, I have been watching an American Studies professor at a private college in Western New York write about artificial intelligence with the intensity of someone watching a storm approach. He writes about it often. Sometimes several times a week.
I understand the concern. Anyone who has taught writing, literature, history or philosophy has probably had the same uneasy thought: What happens to an assignment when a student can hand it to a machine and receive something that looks finished in seconds?
Generative AI can invent sources, produce confident nonsense. It can smooth away the awkwardness that sometimes makes student writing recognizable as student writing. And most troublingly, it can allow a student to avoid the difficult part of education: sitting with something they don’t understand yet and working their way through it.
The professor I have been reading has proposed a straightforward institutional response. Universities, he argues, should generally prohibit student use of AI under existing academic honesty policies, while allowing individual professors to make exceptions when they have a good pedagogical reason for doing so.
I don’t think that is a foolish position. Universities shouldn’t dump a rapidly changing technology on faculty and essentially say, “Good luck.
Professors need guidance. Students need clear expectations. Institutions need to think seriously about assessment, academic honesty and what they actually mean when they say they want students to learn.
But one thing the professor wrote got me wondering.
“AI, we’re told, will make learning more efficient. But is that what we dream of for our kids — an efficient childhood? To write an efficient story, sing an efficient song?”
It is a beautiful question. And it made me think about how often education has had this conversation before. Long before ChatGPT, people worried that new technologies would make human beings intellectually weaker.
In Plato’s Phaedrus, Socrates worries about writing itself. If people can put knowledge outside their own minds, he suggests, they may begin to possess the appearance of wisdom without actually possessing wisdom.
His warning sounds remarkably familiar:
“This discovery of yours will create forgetfulness in the learners’ souls, because they will not use their memories.”
There is something almost funny about this now. Western education spent the next 2,500 years building itself around writing. The printing press brought another round of anxiety. Suddenly there could be too many books, too much information and too much reading. The novel was periodically accused of corrupting young people. Newspapers were criticized for sensationalism and distraction.
Then came radio, film and television.
Each seemed to offer the possibility of putting knowledge directly into the minds of millions of people. Each raised a familiar question: If technology can deliver information so easily, what happens to the teacher?
Apparently, the teacher survives.
That doesn’t mean the fears were always foolish. Technology really does change us. It changes what we practice, what we remember and what we expect from ourselves.
The calculator is a good example. When calculators became common in classrooms, some educators worried that students would stop learning arithmetic. That concern wasn’t entirely wrong. There are things a student should be able to calculate without reaching for a device.
But the calculator also forced educators to ask a better question. if a machine can perform a calculation instantly, what part of mathematics actually belongs to the human being?
That is where I think we are with AI.
If a machine can produce a five-page essay in thirty seconds, we need to ask what the five-page essay was supposed to demonstrate in the first place.
Was it simply the ability to produce five pages of competent prose because the assignment said five pages?
AI has exposed something uncomfortable about higher education: often we aren’t entirely sure whether our assignments are measuring learning or compliance with the rituals of learning.
That doesn’t make AI harmless.
If anything, this technology is more difficult than many of its predecessors because it doesn’t merely deliver information. It can imitate some of the intellectual products we ask students to make. It can write the essay, summarize the book. It can generate the code, and it can construct the argument.
That is a real problem for conventional assessment. But a problem is not the same thing as the end of something. The Internet didn’t eliminate universities. Google didn’t eliminate libraries. Wikipedia didn’t eliminate knowledge. Smartphones didn’t eliminate conversation.
They changed what we had to do. AI will change it too. Students will have to talk more about their work instead of simply submitting it. Drafts and research notes may matter more. Oral examinations may make a comeback. Professors may ask students to explain why they made a particular argument rather than simply reading the finished argument.
Some assignments may disappear.
The purpose of education is not to preserve the five-page essay, the research paper or any other particular academic ritual. Those things are means.
The end is supposed to be a person who can think.
And this brings me back to the professor’s question about efficiency. No one wants an efficient childhood. We don’t want efficient friendship, or efficient music. We don’t want efficient love.
We don’t read a poem because it is the quickest way to acquire information. We don’t walk through the woods because walking is the most efficient way to get somewhere.
Some things are worth doing because of what happens while we’re doing them. Education is one of those things.
But I also think we should be careful about making efficiency itself the villain. Technology has always made some forms of human work easier. The printing press made books easier to reproduce. The calculator made arithmetic easier. The word processor made revision easier. Search engines made information easier to find.
Sometimes that is good.
The question isn’t whether technology saves us effort. The question is what we do with the effort it saves.
That, to me, is the more interesting question about AI.
If students no longer have to spend hours doing certain mechanical tasks, we should ask them to spend those hours doing something machines are much worse at: discussing an idea with another person, defending an interpretation, paying attention to a particular place, reading a difficult book slowly, making something that doesn’t quite work and trying again.
Maybe the answer isn’t to make education harder just because it used to be harder.
Maybe the answer is to make it deeper.
There is another reason I hesitate to describe AI as the beginning of the end of education. Employers are still looking for people who can do remarkably old-fashioned things.
Read carefully. Write clearly. Recognize a bad argument. Distinguish evidence from assertion. Listen to another person. Make a judgment when the answer isn’t obvious.
Those skills aren’t made less important by AI.
A student who cannot tell whether an AI-generated argument makes sense is in a vulnerable position. So is a student who cannot recognize a fabricated citation or explain an idea without the machine’s help.
The better AI becomes at sounding intelligent, the more important it may be for human beings to understand what intelligence actually looks like.
That is why I don’t think the answer is simply to keep AI outside the classroom.
There will certainly be courses and assignments where students should not use it. There should be places where the point is to see what a student can do unaided.
But students are going to graduate into a world in which AI is part of ordinary professional life. Pretending otherwise doesn’t prepare them for that world.
The university has to teach students how to use these systems without allowing the systems to do their thinking for them.
That is a much harder assignment.
It is also, perhaps, a more interesting one.
And here I think the professor is absolutely right: AI is not merely an individual faculty problem.
It is an institutional problem.
Professors shouldn’t have to become software ethicists, plagiarism investigators and technology-policy experts on top of everything else they already do.
Universities need to decide what they value and then build policies around those values.
But a policy doesn’t have to be a prohibition.
A blanket ban can protect some traditional assignments. It can also create the fiction that students will encounter AI only when a professor gives them permission.
They won’t.
The question is what happens after graduation.
That is why I keep coming back to history.
Writing changed education. Printing changed education. The novel changed education. Radio, film and television changed it too. So did calculators, computers and eventually the Internet.
None of these technologies left the university untouched. But none eliminated the basic reason we gather human beings together to learn.
We still need teachers.
Not because teachers can compete with machines at producing information. They can’t.
We need teachers because education has never been only about information.A good teacher helps a student notice something they had overlooked. They challenge a lazy assumption. They recognize when a student has finally understood something. They ask a question that sends the conversation somewhere unexpected.
Sometimes they simply sit with a difficult question long enough for the student to become comfortable not knowing the answer yet.
A machine can assist with some of this. It cannot make the human relationship disappear.
Marshall McLuhan put the problem well:
“We shape our tools and thereafter they shape us.”
That cuts in both directions. AI will shape the university. But the university will also shape the way students understand and use AI. That is the opportunity hidden inside all this anxiety. AI is forcing higher education to say out loud some things it should have been saying all along: What is education for? What should a person actually know? What should a person be able to do without assistance? What kinds of struggle are worth preserving? And which forms of academic struggle have we preserved simply because nobody stopped to ask whether they still make sense?
I don’t think those questions can be answered with either panic or enthusiasm.
They require judgment. The professor I have been reading is right to worry. I am right to worry, too. Anyone who cares about education should be worried enough to pay attention.
But there is a difference between taking a technology seriously and assuming it will destroy the institution it enters.
Education has been absorbing disruptive technologies for a very long time. It has changed. It has adapted. Sometimes badly. Sometimes brilliantly.
I suspect AI will be the same.
The real danger may not be that artificial intelligence makes human education obsolete.
It may be that, while arguing about the machine, we forget to ask the most important question:
What is it that only human education can do?