The silent despair of teachers in the age of artificial intelligence

Artificial intelligence has arrived in classrooms, and with it came a feeling that many teachers are still trying to name.

It is not exactly anger.

It is not resistance to technology.

It is something closer to grief.

For decades, the heart of teaching lived in a very specific place: that moment when a student stumbles over a difficult idea, wrestles with it, and finally reaches some understanding that is genuinely their own.

That process — messy, time-consuming, sometimes frustrating — was always the point.

And now it is disappearing.

The arrival of large language models like ChatGPT has transformed the relationship between students and learning itself in a way that neither optimists nor pessimists managed to predict accurately.

Assignments come in well written, well structured, with coherent arguments.

But something essential got lost along the way.

This article dives into that real tension, told by people who are inside classrooms every day, facing a question with no easy answer: what is education when the final product can be generated by a machine? 🤔

The lament of Arjuna and the mission of teaching

There is a passage in the Bhagavad Gita, a sacred Hindu text, where the warrior Arjuna surveys the battlefield of Kurukshetra and realizes he will have to fight against friends and family. Overcome with despair, he questions what his duty is in such an extraordinary situation. Krishna then responds with a teaching that has echoed across centuries: do your duty without attachment to the outcome.

The author of the original The New Yorker article says he discovered this story in a Hinduism class in college. He was not exactly a dedicated student — he skipped classes often and his grades were nothing to brag about. But a patient professor, the kind you find in religion departments at universities in the northeastern United States, explained the text in a way that was accessible and generous. And that explanation stuck. Twenty-five years later, the author still thinks about Arjuna’s despair and Krishna’s command.

This is exactly the kind of experience at the center of the debate about artificial intelligence in education. The concern is not just about the future of the diploma or the absurd cost of American universities. It is about that instant when a nineteen-year-old is forced to read something they would never have picked up on their own and, guided by a professor, arrives at a discovery that changes how they see the world. How do you put a price on that? How do you protect that experience when AI can deliver ready-made answers in seconds?

What actually changed in the classroom

For a long time, teachers evaluated students by the process just as much as by the result. An essay full of crossed-out lines, a problem set with the reasoning spelled out, a poorly constructed argument that still showed someone was trying to think — all of that said something. It said there was a developing brain behind that paper. Today, when a student turns in a flawless text on a complex topic, the first reaction of many educators is not admiration. It is suspicion. And that reversal, all by itself, is already a sign that something deep has shifted in the dynamic between those who teach and those who learn.

Artificial intelligence, especially large language models, did not arrive as an optional tool sitting in the corner of the room. It showed up as a constant presence in the daily routine of anyone who studies. Recent surveys indicate that a significant share of college students already use AI tools to write, summarize, or structure academic assignments on a regular basis. We are not talking about isolated cases or occasional cheating. We are talking about a behavior that became normalized so quickly that many young people do not even see it as problematic — after all, if the tool is available, why not use it?

Jane Sloan Peters, a professor of religious studies at the University of Mount Saint Vincent, lived this firsthand. She created a course called Letters from Prison, about what people throughout history have been able to endure because of their faith. The highlight of the course was the moment when students tried to synthesize a general theme from everything they had read. When she first taught this class four years ago, students struggled to reach their own conclusions. Through brainstorming and revision, they eventually arrived at an understanding that was personal and showed real engagement with the texts.

Last year, everything changed. None of the sixty students seemed to have any difficulty with the assignment. Peters received elegant summaries, the kind you would find on the back cover of a book, and perfectly generic course themes that somehow addressed everything without really saying anything. The suspicion that AI was behind it led her to add handwritten brainstorming steps to the course, hoping to make it AI-proof. But when she presented these new expectations to the class, something unexpected happened. A wave of sadness washed over her and she got emotional in front of her students. Peters told them that, before AI, students worked hard to arrive at their own ideas. She helped, they struggled, but they got to something that was genuinely theirs. And that was not happening anymore.

The despair is not about technology

Let us be clear: the despair many educators report is not technophobia. It is not that old story of teachers resisting computers in the 90s or Google in the 2000s. It is something different and, in a way, more unsettling. Because this time, the technology in question does not just make information easier to access. It replaces the act of thinking, at least in its most visible form. And that poses a philosophical challenge before it is even a pedagogical one.

Peters and other professors interviewed do not blame the students, and they make clear that not all students welcome the changes brought by this new technology. Some students even express frustration with professors who simply allow unrestricted use of AI. The line Peters heard from students sums up the feeling well: why don’t you expect more from us than that? Even those who use the tool, on some level, want to be held to a higher standard.

The problem, as Peters identifies it, is that AI arrived on top of a transactional model of education that had already been in place for a long time. Students are told they are in school to get a degree — a degree that comes with a steep price tag and, for many, a significant debt burden. They are evaluated by the work they turn in. And because AI allows them to submit material that, on the surface, looks reasonably good, many do not understand why it is such a big deal when they cannot explain what they generated. Peters admitted she feels waves of relief when she sees spelling mistakes and clunky grammar in assignments — because it means the student is actually thinking on their own. 😔

Voices from the front lines: how AI is transforming teaching in the United States

The original The New Yorker article features testimonies from twelve professors at American universities, each offering their perspective on how artificial intelligence has altered their work. What emerges from these accounts is a complex portrait, sometimes contradictory, but always charged with emotion.

The fear of a career coming to an end

Susanna F. Boxall, a philosophy professor at California State University, Chico, is blunt: if her job still exists when she turns fifty, she will consider herself lucky. She is forty-five now. For Boxall, the combination of AI with the so-called demographic cliff — the population decline that is reducing the number of potential students — is devastating higher education. Major research universities may survive, but smaller institutions like hers will shrink or disappear.

Boxall describes post-AI online classes as a simulacrum of education: students pretend they are learning and she pretends she is teaching something. In-person classes still maintain some rigor, but she can no longer assign homework because somewhere between seventy and one hundred percent of students will use AI. In a recent semester, she managed to do an oral final exam with a class of eleven students, reserving a room for six straight hours. Scaling that to classes of a hundred and fifty students is simply impossible.

The shift in computer science education

Kevin Sun, a computer science professor at the University of North Carolina at Chapel Hill, is also pessimistic. The most obvious change in his teaching has been eliminating difficult homework problems, which used to be an essential component of the course. He has tried using social pressure as motivation, with group quizzes and classroom presentations, but acknowledges there are limits to what an individual professor can do against systemic forces like grade inflation, the job market, and student evaluations of professors.

On the other hand, Sun recognizes benefits: AI has helped him create syllabi, lesson plans, and exams. He has also designed activities where students evaluate AI-generated code, which may be correct or contain subtle errors. In computer science, AI has shifted the emphasis from writing code to evaluating code — and training students for that is the new challenge.

The reinvention of sociology

Daniel Silver, a sociology professor at the University of Toronto, reports that AI has fundamentally changed how he teaches. He invested time creating multi-agent simulations where students build representations of theories from thinkers like Adam Smith and Max Weber, experimenting with them. The commitment was enormous for everyone involved, but the best final projects showed more creativity and intellectual effort than the typical second-year essay.

Silver also adopted a direct approach when he detects thoughtless use of AI. He calls students in for individual conversations lasting thirty to forty-five minutes, gives a zero on the assignment, and allows a redo after the conversation. For him, the most important thing is that the student feels someone — especially a professor — is paying attention to them and to what they produce. Something that, unfortunately, is rare at larger universities.

He also shows students what he calls replacement-level work, inspired by the sports analytics concept of wins above replacement. These are variations of AI-generated responses to assignments. Students can clearly see how they all look the same. Those are C-level answers at best, and students understand they need to produce something better than what the machine would do.

The drama of online teaching

Elizabeth Strom, a professor at the School of Public Affairs at the University of South Florida, teaches many fully online classes and admits there is no way to prevent AI use in that format. Most students do not know her personally, nor do they know their classmates, so the social norms that reinforce doing your own work simply do not exist. She tries to design activities that make it harder to skip the reading — she asks for citations, page numbers, personal opinions — but acknowledges it is possible to get around those requirements. And with fifty students, she does not want to spend her time playing detective over who actually wrote which assignment.

The theater professor who became a plagiarism cop

Neal Hebert, a theater professor at Grambling State University, may have one of the most striking accounts. He does not analyze plays as literature. Instead, he asks students to imagine them as the bones of a human body, which only come alive when performed on stage. The assignments he gives call for personal reflection, like choosing two words that describe the physical world of a play and explaining why.

When he read the assignments on August Wilson’s Fences, the disappointment was immediate. Out of forty students, the majority chose similar words, phrases, and concepts, all written in that unmistakable ChatGPT style. Hebert compares the result to elevator music, but in words.

The solution? He started selecting plays too obscure for ChatGPT to have information about. When AI is used on those assignments, it hallucinates — invents characters, plot lines, simply fabricates information. But Hebert resents the fact that he went from being a creative collaborator to being a plagiarism cop. He wanted to be the kind of professor that his own professors were for him.

Honors students outsourcing their own thoughts

Beth Ritter-Conn, a professor of religion at Belmont University, describes the turning point: when she discovered that honors program students were using AI to write reflection journals. The only task was to say what they were thinking inside their own heads. There was no right or wrong answer. And still, some outsourced that task to a machine. For Ritter-Conn, it is impossible to give honest feedback when the work is not honest. And it is impossible to help someone think about how they want to exist in the world if that person is not using their own brain to say where they are.

Financial fraud and ghost students

David Song, a professor of Asian American studies at East Los Angeles College, raises a troubling dimension that seems more specific to community colleges. Beyond the routine use of AI by real students, he reports cases of financial aid fraud, where fake students enroll in courses and use AI to complete assignments. Song remembers noticing something odd when, in his introduction to Asian American history course — at a college with a predominantly Latino and Asian student body — three or four students with generic names showed up writing posts about how they had already taken other Asian American studies courses. For him, it just did not add up.

Hope still persists in Houston

Jeremiah Croster, an English instructor at Houston Community College, works at one of the most international community colleges in the United States, with students from Africa, Asia, Mexico, Central America, and even Kazakhstan. He estimates that between fifty and sixty percent of his in-person students and between eighty and ninety percent of his online students were using AI during the spring 2024 semester.

Tools we use daily

Translation

Text Inspection & Clipping

Social Media

Search & Research

Recruitment

Programming

Productivity & Organization

Operations

Marketing & Sales

Finance & Investment

Education

Design & Media

Data & Analytics

Content & Writing

Automation

Croster switched to handwritten exams in blue books, replaced written responses with videos where students talk about the topics, and started deducting points for common ChatGPT writing tropes, like three-item lists and excessive use of adjectives. In his assessment, ChatGPT is still a bad writer, incapable of handling the existential risk involved in crafting a truly compelling thesis.

A colleague of Croster’s used to say there are two approaches to college teaching: one helps students get an education and the other helps them get a degree. The degree approach was already winning before AI. Now that it has arrived, the education part is starting to feel like something someone will write about in a history book. Or maybe AI will do that.

What students are losing without realizing it

Students who grow up delegating cognitive tasks to artificial intelligence face a risk that does not show up in grades or evaluations yet, but will at some point. The ability to reason under pressure, to build arguments without external support, to sit with the frustration of not understanding something right away — all of that develops precisely in the difficult situations that AI eliminates so efficiently. It is like using GPS for every single trip and then discovering you have completely lost the ability to orient yourself in space. The shortcut works, but the muscle atrophies.

There is also a question of intellectual identity at stake. Part of what education has always done is help young people discover how they think, what they value, how they organize their ideas about the world. A text written by a student, even an imperfect one, carries the marks of that process of self-discovery. A text generated by AI, no matter how sophisticated, is a statistical simulation of how someone might think about that topic. It is not the student’s voice. It is a well-calibrated approximation, but one empty of personal experience. And when those texts pile up over years of schooling, what remains? What intellectual foundation does that young person carry with them when they leave school or college?

Lauren Aulet, a professor of psychological and brain sciences at the University of Massachusetts Amherst, captures the tension well. For her, AI has expanded what is possible to do scientifically, lowering the cost of experimenting with things computationally. But the same tool that is genuinely useful for research can be destabilizing for education. Aulet’s line is precise: AI helps me code does not directly translate to AI is good for universities. 🧠

The bird is in our hands

Jane Sloan Peters closes her testimony with a reference to Toni Morrison’s Nobel lecture in 1993. Morrison retells a fable in which young people ask a wise old woman whether the bird they are hiding behind their backs is alive or dead. She answers: it is in your hands. The bird is language.

In Morrison’s retelling, the young people are not fools or pranksters. Behind their provocation is a genuine plea: that the older generation take them seriously, that it acknowledge the ways it has failed them. Peters sees a direct parallel with the current situation. Students use AI even when they are uneasy about what the technology will mean for their lives. And on some level, they are looking to their professors for guidance on how to use it — and, more than that, for reassurance that their words still matter. That they still matter.

Possible paths through the chaos

Some educators and institutions are already experimenting with approaches that try to coexist with artificial intelligence without giving up what makes education meaningful. Oral assessments have made a comeback at universities that had abandoned them years ago. Multi-stage writing processes, with drafts, revisions, and justifications for each choice, are being reintroduced as a way to make reasoning visible. Some schools are betting on classroom discussions, in-person collaborative projects, and live presentations as ways to evaluate what a student has actually internalized.

Auyon Siddiq, a professor at UCLA’s Anderson School of Management, represents the other end of the spectrum. He encourages the use of AI for technical details, like coding, so students can focus on the concepts. His class took a fully AI-open exam as an experiment, with the only restriction being that students could not photograph the test with their phones. The class average still came in at seventy-five percent, because the students who were truly lost were not saved by AI. For Siddiq, the tool freed up mental space to think more about class structure, pacing, anecdotes, and case studies, instead of being consumed by the mechanics of creating slides and homework.

The idea behind these changes, in all their various forms, is simple but powerful: if AI can produce the final product, then the final product can no longer be the only criterion for evaluation. The process needs to return to the center. And that requires a deep reconfiguration not just of teaching methods, but of the entire school culture around what it means to learn well and what it means to turn in good work. This kind of change does not happen from one semester to the next. It takes time, experimentation, and a considerable amount of patience from everyone involved.

What is clear is that the despair of teachers, when channeled productively, can be the engine of a necessary transformation in education. Not a transformation that rejects technology, but one that insists on asking what it is for and what it cannot replace. Artificial intelligence will keep evolving — that is not up for debate. The question is whether educational institutions will manage to evolve alongside it, keeping at the center what has always been irreplaceable: the experience of a human being learning to think for themselves. ✨