Generative AI use in academic research neither compulsory nor unobjectionable

31 Oct, 2026

[The usual shorthand caveat, stated mostly for purposes of making sense in future eras: in this post I use "generative AI" to mean the large-language-model-based products currently being marketed to us as such by companies like OpenAI and Anthropic, particularly for endeavors related to academic work, such as programming, formalizing mathematics, and writing papers and grant proposals.]

I am a tenured faculty member at a university whose upper level administration is aggressively pushing and investing in generative AI usage in teaching and research. Further, many of my colleagues are on board with this and actively arguing within my professional communities and research area that we should be using these products as much as possible. I am opposed to this.

I don't use generative AI in my teaching, research, or personal life if I can help it. I think these products doing immense harm to basically everything and everyone I care about and that the harms outweigh the benefits.

There are a lot of reasons for this, but I am not really trying to focus on what I personally do or don't do, or claim that I will never change my mind. What I'm trying to do here is cut through the din of sales representatives insisting that not only are we obligated to use and engage with it, but that we should be eager to adopt it into our existing practice because of all the obvious upsides, and conveniently gaslighting you by failing to acknowledge the downsides. My counterthesis is that no, you don't have to; no, I don't think you'll necessarily be worse off in the ways you're afraid of; and yes, I think all the downsides are still good reasons to stand firm in your refusal, cautious moderation, hesitation, uncertainty, or wherever you happen to be.

Here are some reasons.

I don't see problems with how we were doing things before that are made better by these tools. They might be done faster, with more output per unit of time, but I don't see that as inherently better.
Indeed, it is often demonstrably worse, resulting in high-volume, sloppy work that DDOSes the attention of everyone now obligated to review it (be it papers, code, or proofs). It turns into an adversarial endeavor what used to be an earnest, if heated, battle of ideas between parties we could on some level take for granted were acting in good faith to publish true things that advance human knowledge; the punishment for breaching this social contract would be massive reputational damage. Peer review and open source, two processes that used to represent healthy and important alternatives to commercialization via capital, are now choking to death.
Of course, there are new problems that people are deciding to work on because these tools have either introduced them or made them more viable. But I like using my own judgment to choose what problems to work on, and I am happy with the ones I've been choosing lately.
Academics have a responsibility to produce and share knowledge we believe in based on evidence derived from rigorous methods, regardless of whether that knowledge makes other people happy or want to give us money. If we let manufactured hype not only sell us a story of how specific corporate products will be useful to us, but also demand that we turn into sales representatives for those same products, we cannot and should not be taken seriously.

Some worry about damaging their reputations in circles that value quantitative, surface-level measures of success. I worry about damaging my reputation among the people I look up to and value as my peers, who care about things like careful thinking; thoughtfully-designed abstractions; fundamental principles; human understanding, curiosity, insight, and skill development; moral character; and a willingness to go slowly and get things right, even at the cost of perceived productivity and prestige.

A lot of people are leaving, or considering leaving, their professions and fields because of how this "moment" has steamrollered everything they cared about. This last one is actually in my view the most important. It includes the students we teach and advise, who are scared for their futures. They need to figure out what they will do with their degrees, and they will leave if they don't like the options. It includes my generational cohort, many of whom have been subject to layoffs or optionally left their jobs because they didn't want to participate in what they see as being forced on them. And it also includes academic and industry veterans who were already near retirement and decided sooner would be better than later. (I count at least three of my own mentors, formative to my career, among them.) If nothing else, we should be putting some energy into building (or supporting people who are building) literally anything else where folks opting out might find their considerable talents valued.

I've resisted writing my thoughts about generative AI in longform for a long time, because in addition to feeling pressured to use it, I also feel pressure to talk about it, constantly, in every professional discussion. I do not like caving to pressure I see as coming downstream from actors who wish to damage the livelihoods of people I care about.

But it's increasingly clear that those of us who don't feel compelled to use, or seriously engage with the prospect of using, generative AI, have a responsibility (to one another; to the next generation) to speak up about it. Not because I think it's especially likely that we'll change the tides, but because, among other things, an emergent effect of these products is alienation, both from our work and from each other. I speak often to people who, in quiet voices in the wake of much louder ones, confide in me that they hate what is happening, but they feel totally isolated in that feeling, and that if they don't join the bandwagon, they will be socially outcast, bereft of not just career prospects but of a professional community who respects them and their decisions. Note the irony of this fear in contrast with the fact that I have more of these conversations with new people almost every day.

It would seem that the only way to convey that folks are not alone is to speak up about it. Tenured academics in particular, a group I am fortunate to be in, hold particular privilege to do so. So I feel a responsibility to declare my values and create space for others who share them. Those spaces may not be well-funded, featured in university public relations materials, or beloved by a majority of colleagues. But they will be important to the people who need them, and for me, that's enough.

Misc things I read on this topic that influence my thinking, and generally cover other concerns I share but didn't talk about:

"An anecdote against slop artifacts", Markus de Madeiros https://www.markusde.ca/pages/noslop.html

"Why I don't use generative AI", Jonathan Chan https://ionathan.ch/2026/03/18/LLMs.html

"Against Vibes", parts 1 and 2, William J. Bowman https://www.williamjbowman.com/blog/2026/03/05/against-vibes-when-is-a-generative-model-useful/

"Early career researcher responses to AI in math" https://arxiv.org/abs/2610.40584