Humanities departments seem to be in perpetual crisis. Fewer students are enrolling in them. The Trump administration is cutting their funding. Smartphones and social media are hastening the collapse of reading and attention spans, even among students at elite schools. Americans are becoming more skeptical about the economic value of any four-year degree, let alone one in comparative literature.
In answer to these and other challenges, many colleges are trying to make the humanities “relevant.” Some are accommodating reduced attention spans by assigning excerpts rather than books. Others are responding to financial anxieties by restructuring departments to emphasize their practicality (if they aren’t eliminating programs altogether). But such adaptations and compromises only exacerbate perhaps the most insidious threat the humanities face, and one that’s not often discussed.
As a humanities professor myself, the biggest danger I see to the discipline is the growing perception, fueled by the ubiquity of large language models, that knowledge is cheap—a resource whose procurement ought to be easy and frictionless. The humanities, which value rigorous inquiry for its own sake, will always be at odds with a world that thinks this way; that’s why relevance is a futile goal. For humanities departments to continue to matter, they must challenge the modern world rather than accommodate it. Indeed, the most useful lesson the humanities have to offer today is a profoundly countercultural one: Difficulty is good, an end in its own right.
For the past several years, I’ve had the pleasantly vague title “visiting professor of humanities” at Bard College, a small liberal-arts school in New York’s Hudson Valley. Bard has given me a remit as simple as it is generous: to teach books and ideas I think are important. Every November, I submit course descriptions for two spring seminars—this year, one on Albert Camus and his influences, the other exploring the idea of the American dream through Black writers such as Frederick Douglass and James Baldwin. Within a few days of the courses being posted, prospective students start writing me to say how eager they are to immerse themselves in the texts. I’ve learned to relish their zeal, because I know it won’t last.
When they start my class, many of my bright, self-selecting students appear to be unacquainted with the difficulty of close reading. By the end of the semester, only a fraction seem to have gotten through the texts and writing assignments without outsourcing at least some of their work to AI. In my course on Camus, most students will be able to remember the philosopher’s famous injunction to imagine Sisyphus happy, but few will demonstrate mastery of the abstruse train of thought that led him to it. Not many are fully willing to try.
I began teaching in early 2023, two months after OpenAI released ChatGPT. It could generate some amusing parlor tricks—a personal favorite was interweaving Shakespearean sonnets with mid-’90s rap—but it wasn’t good at much else. During my first semester, one or two students turned in writing that featured AI’s distinct blend of fluency and superficiality, which was easy to detect.
Chatbots look very different today. As the technology has become more sophisticated, more of my students have tried to pass off AI-generated writing as their own. The craftier ones will use chatbots to come up with phrases or insights that they shape into their own prose. The work this yields is usually competent, making AI’s influence difficult to spot, but not exceptional. If I have suspicions about a student’s take-home essay—maybe it looks nothing like their in-class writing assignments—I run it through AI detectors. These are far from perfect, of course. But when they indicate the use of AI, I confront the student, and he or she almost always confesses. Still, in just three years, ChatGPT and its competitors have rendered take-home essays—what I consider the central exercise of humanistic learning—nearly useless to assign and almost impossible to assess.
More recently I’ve come to suspect that, in addition to using LLMs to ghostwrite papers, some of my students are relying on them to prepare for in-class discussion. At any rate, their contributions are getting blander and more interchangeable, less daring; eccentric or original observations are becoming rarer. If I’m right that AI is furnishing my students with talking points, then it has almost entirely eliminated the possibility that they arrive at some transformative insight on their own, which is what makes wrestling with words and ideas so joyful and fruitful in the first place.
Probing a text can be enjoyable but also tiring, even borderline painful. That’s good. Exhausting our mental faculties, such as through deep reading or effortful writing, is what makes them more potent. Physical exercise works the same way. AI, by contrast, promises knowledge without effort, just as many people see in GLP-1 drugs the possibility of weight loss without willpower. Although both have legitimate uses, their widespread adoption has diminished our capacity to appreciate, let alone endure, the sustained and challenging work required to flourish beyond the level of simple appearance. Only through difficulty do we improve our powers of thought and perception, which we carry with us in every endeavor. This is the true source of the humanities’ relevance.
Camus’s great realization was that, in a meaningless world, we create our own meaning and quality through willed struggle—a lesson that AI threatens to obscure but the humanities are uniquely poised to teach. Sisyphus is assigned to roll his rock for eternity, Camus writes. Yet he can still be happy so long as, each time he comes to the bottom of the hill, he’s the one who chooses to turn around and rise back up.
