The Students Who Won’t Use AI
A meaningful minority of college students refuse to use AI—and what they’re telling us about the real problems in our classrooms.
In my last post, I argued that “I’m against AI” isn’t one position but three—structural, cognitive, and pedagogical—and that the conversation stalls because we rarely bother to sort them. I wrote it thinking about faculty, but a piece in this week’s Chronicle made me realize I should have been thinking about students, too.
Beth McMurtrie’s April 16 Teaching newsletter reports something the dominant AI-in-higher-ed narrative tends to hide: a meaningful minority of college students are low-to-no AI users. A Gallup-Lumina survey of nearly 4,000 students found that 13% had never used AI in their coursework, and another 19% used it only infrequently. The direction of the data is also striking: over the past year, Gen Z sentiment toward AI has moved against the tool, not with it—with excitement down 14 points, hopefulness down 9, and anger up 9. A majority of Gen Zers now say AI designed to complete tasks will make learning more difficult rather than less. The arc is not bending toward uncritical embrace.
I’ll add a data point the surveys don’t capture. In my own house, I’ve watched my high-school kids labor over a piece of writing and then—knowing full well that they wrote every word—run the finished draft through an AI checker to make sure it doesn’t trigger even the appearance of AI use. “Teachers are going to run it through a checker anyway,” they tell me, “and if it triggers anything they’re going to make you redo the whole thing. Why risk it?” That’s AI being used to police AI, and it’s doing its own part to erode the trust on which the classroom depends. This is not the behavior of an entire generation sleepwalking into the technology, and arguably points less to problems with student use than to problems with assignment design that teachers must figure out.
That’s AI being used to police AI, and it’s doing its own part to erode the trust on which the classroom depends.
I want to sit with that for a moment, because it’s a more hopeful—and more complex—story than the one we usually hear. A significant portion of our students are arriving on campus with a clear sense of what they value in their learning. When they push back against AI-generated work, it’s rarely about being contrarian; it’s an active choice to protect the “labor of thinking” for themselves. They aren’t outliers; they’re students who have identified a boundary they value, and they’re standing by it.
And when they articulate why they refuse AI, they don’t sound like the hype cycle. They sound, to my ear at least, a lot like the faculty I’ve talked to. Or, as McMurtrie puts it bluntly: “Young people aren’t that different from their teachers and professors in this regard.” If she’s right—and the numbers suggest she is—the thoughtful middle isn’t just where we live. It’s where a substantial part of our classrooms live too.
McMurtrie cites reporting by her colleague Emmy Martin, whose “When AI Use Makes You Uncool” describes students who treat AI use in class as “a character test. A marker of how seriously a student takes the labor of thinking.” They enforce the norm through jokes, side-eye, and anonymous posts mocking “AI slop” and suggesting that if you can’t write an essay without chat, you might not deserve to be there.
Look at the three clusters of objections I discussed in the last post and something interesting happens: the structural and cognitive objections are present, but refusal—by those who don’t want to touch AI—is a pedagogical objection, and it is being enforced informally by peers. A chunk of our students, in other words, have already done for themselves the sorting work I was trying to do for faculty in my first post—without a framework, a vocabulary, or anyone assigning it.
That makes me see the pedagogical problem a bit differently. It’s less about trying to stop the AI tidal wave, and more about something harder and more interesting: what would a classroom look like that met all of my students where they actually are, however they feel about AI? The refusers are already doing something the embracers aren’t: recognizing, in real time, when a tool would damage the work. The embracers are already doing something the refusers aren’t: learning to work with a technology that isn’t going anywhere. What would it take to help each group learn what the other has figured out?
This is the best argument I know for why the one conversation we keep avoiding—the one about AI, in our actual classrooms, with our actual students—is the one we most need to be having. College at its best has never been about the delivery of content. The magic and the privilege of higher education is what happens when you put people in a room and ask them to take a hard question seriously together. It is people sitting at a table, thinking out loud, disagreeing in good faith, changing each other’s minds a little. The refuser learns from the embracer what the tool can actually do; the embracer learns from the refuser the actual costs of using it. Neither of those conversations happens on a laptop alone at midnight. We should be having them across a seminar table.
Most of our current AI-policy conversations miss this because they assume a monoculture that doesn’t exist. Ban it, require it, integrate it—each policy move, including syllabus statements about a single class, treats the classroom as a single population with a single problem. It isn’t. And the students themselves, through their own informal sorting, are already telling us so.
One question before I let you go. If a third of your students are already low-to-no AI users and most of the rest are ambivalent about the tools they’re using, the design problem is an assignment problem, not a policy problem. Which of your assignments is doing the work you want it to be doing—and which one, if you’re honest, are you no longer sure about?
Hit reply. I’m listening.
Interested in how I use AI in this newsletter? Read more here.
I’m a senior in college, and I chose to drop out of a creative writing class (one I was very excited to take) after being told I had no choice but to use it to generate outlines for my novella. I don’t want AI to do my thinking for me; I don’t want AI to steal a key piece of my creative process from me. But I was told that the creative writing department was being tasked with including AI in every course and that I “needed” to learn to use it in order to decide whether I wanted to incorporate it in my creative process.
To me, the issue with AI at college is the lack of consent. We have no choice but to participate. I told my professor I had religious objections to using AI (I go to a religious school) and I was quickly dismissed with a pre-written answer that didn’t address my concerns. It feels like intellectual rape. And why am I paying for an education that isn’t teaching me to think or create? My money is just going straight to harmful systems, and it’s so disheartening.
One of the saddest things about AI is the loss of trust that comes with it. Students are afraid of being accused of using it, and teachers are afraid students are using it, and no one can simply read a sentence and assume it was written by another human.