21 The College Essay is Dead. Or Is It?
AI Killed the Essay Star
In December 2022 (so, three years before this book was “written”), novelist and essayist Stephen Marche wrote an article for The Atlantic entitled The College Essay is Dead.
A provocative title, especially to anyone who teaches the “college essay.” It’s a title that suggests a lot of different things – potentially – and most of them don’t seem like they’d be very good for those of us who teach college or for our students. And, if we believe the claim Marche makes in his title, it would seem like there’s little point in carrying on. We should cancel all college essay assignments and give up entirely.
Let’s pause for a moment and look at the picture above. That image was created with Canva, using their AI image generation and this prompt: an image of a mournful college professor who is thinking that there’s no point in teaching anymore as AI can do everything better than his students could ever learn to. So, according to AI, that is what a college professor looks like.
I used that prompt to generate sixteen images of the ‘mournful college professor’ and out of those sixteen images, eleven of them looked remarkably similar. They were all middle aged to actually old white dudes, sitting in an office staring pensively into space or looking as if they might cry. As a middle aged white dude college professor, I can assure you of two things:
- My office looks nothing like that. None of my collogues’ offices do, either.
- I’ve never sat at my desk mourning the future of academia with tears threatening to spill down my cheeks.
The remaining five images were similar to each other in two significant regards: they were all men and they were all Asian men. So, based on the results of four separate instances of running that same AI prompt, I think it’s safe to say that all college professors are men who are either white or Asian.
Or, you know, not.
Why does that matter? What point does an AI generated image make that connects to the question of whether the college essay is dead? Well… Marche – and many other writers and anxiety-riddled college professors – claim that GenAI can now write a serviceable (or better) college-level essay or research paper. They’re not entirely wrong (see next chapter) but they’re also ignoring some of the very valid and obvious arguments against that claim, including the one demonstrated by that image.
GenAI doesn’t always get it right. And how can it? Much like the students those writers discuss, it’s still learning. Unfortunately, much like those students (again), it’s learning primarily comes from the internet and I think we all understand why that might be something of an issue.
(If you don’t, you’ve probably sent money to a foreign prince who needs your money to access a vast fortune they’ll share with you, believe that we need to eliminate Dihydrogen Monoxide from our schools and public parks, or that there was a pedophilia/child trafficking ring run by Bill and Hillary Clinton running under at Washington, DC pizza shop.)
The Canva created image of the mournful professor exemplifies the issue – it’s based on what AI has learned from the internet where stereotypes and biases run rampant. That image probably is what many people would picture in their heads when thinking of a college professor, even if it’s not at all accurate in 2025. And it begs the question: if one of the most popular design sites on the internet is still bogged down by stereotypes and biases, what do you think the chatbots students would use to create their essays are like?
Still, many college instructors across multiple academic disciplines seem to be living in fear of AI and what it will do to education, particularly in terms of academic integrity – our ability to trust that students are doing their own work – which forms the backbone of college-level writing.
Ironies and Inconsistencies
In May and June 2023, textbook publisher Cengage surveyed “approximately 600 U.S. higher education instructors” on their feelings about GenAI. The results were somewhat predictable but also a bit illuminating:
- 84% of the surveyed instructors said that the “most concerning risk of Generative AI” was the threat to academic integrity.
- 76% said that their primary use for GenAI was/would be to determine if students were cheating (i.e. using it to write their essays)
Clearly, many academic professionals feel like GenAI is paving the way to rampant student cheating, with students turning to artificial intelligence to do their work. At the same time, there was another result from the survey worth noting:
- 71% of the respondents said that their primary professional use of GenAI was/would be to “create quizzes or assessments” – meaning the tests, essay questions, and other assignments that students would be graded on. So, you know, turning to artificial intelligence to do their work.
Take a moment to think about that.
There’s nowhere in Marche’s article where he addresses that seeming contradiction – do as I say, not as I do – and, in fact, there’s not much mention of the actual “college essay” in particular. It’s much more a commentary on the debate between the humanities and the sciences/technologies and their respective values in education. But, the article does provide a unique starting point for a discussion about the academic essay – a discussion with GenAI.
GenChat with Gemini
I provided Google’s Gemini AI with the following prompt:
Summarize the article “The College Essay is Dead” by Stephen Marche. Focus on providing a list of his main reasons why Generative AI is harmful to academic writing and also a list of his proposed solutions to the problem.
This is Gemini’s response:
Marche’s primary “harm” isn’t just cheating; it is the fundamental “breaking” of the feedback loop between student and teacher. His main reasons include:
- Obsolescence of the “Basic” Essay: AI can now produce a “B+” caliber essay—clean prose, logical structure, and coherent arguments—on almost any standard prompt. This makes the standard undergraduate essay assignment “dead” because it no longer reliably measures a student’s effort or understanding.
- The Invisibility of Detection: Unlike traditional plagiarism, which can be traced back to a source, AI-generated text is original in its phrasing. This creates a “cat-and-mouse” game that professors cannot win, leading to a breakdown of trust in the classroom.
- Devaluation of the Humanities: Marche argues that if the humanities continue to rely on the essay as their sole metric of intelligence, and that metric can be automated, the entire field appears “irrelevant and self-consuming.” He warns that the humanities are “under-theorizing” the impact of tech.
- Loss of the “Human Middleman”: Academic writing often involves synthesizing information. When a machine can synthesize information better and faster than a student, the “labor” of writing—which was previously equated with “learning”—is removed, leaving a void where intellectual development used to happen.
Proposed Solutions to the Problem
Marche does not suggest banning AI; instead, he argues that academia must fundamentally restructure itself. His solutions include:
- A “Total Overhaul” of Assessment: Moving away from the take-home essay as the “gold standard.” He suggests returning to older methods of evaluation, such as:
- Oral Exams: Testing students’ knowledge through direct conversation to ensure they understand the material.
- In-Class Hand-Written Essays: Removing technology from the writing process to ensure the work is original.
- Interdisciplinary Collaboration: Marche argues for a “joining of cultures” between the humanities and the sciences. He believes humanists must stop ignoring technology and start working with computer scientists to understand and shape these tools.
- Redefining the Value of Writing: Educators should shift from grading the product (the finished essay) to valuing the process and the “human” elements of voice, style, and history that AI cannot yet fully replicate.
- Embracing AI as a Scholarly Tool: Ultimately, he suggests that Ph.D. students and researchers will use AI to handle the “drudgery” of data synthesis, allowing human scholars to focus on higher-level original insights and creative connections.
The response from Gemini features two elements of note:
- The points Gemini raises are far more about the college essay than almost anything Marche wrote in his article.
- Virtually every point Gemini claims Marche makes in his article is completely made up. He didn’t say any of it.
So, what does all of this mean? Is the college essay dead? Is it on life support and we need to give it CPR or does it have a DNR? Should we be paying more attention to the human writer who gives his article a title that’s about something the article itself really isn’t or should we listen to the GenAI that made a bunch of stuff up but at least made up stuff that’s on topic?
