On Writing and AI

The sad thing about chatgpt killing essays is that essay writing is a very effective way to learn.

I remember the main ideas of every research essay I wrote in undergrad but only a fraction of what I was tested on.

— T. Greer (@Scholars_Stage) September 6, 2024

From an essay on Inside Higher Ed, writer and professor John Warner wrote:

“The difficulty of writing is the point. If students don’t find the writing hard, something has gone wrong.”

As someone who writes for a career, I am of course conflicted about ChatGPT. It is great at giving me a first draft similar to something I would have taken days to write. Then I just have to edit and revise and now I have something that would have taken me much longer to produce alone and is probably better.

Over the years I have gotten much better at editing other people’s work to improve their writing. Like most people, I’m not good at editing my own work and with ChatGPT I don’t have to. But I feel like the reason I have gotten better at editing is from all my years of writing and revising and getting feedback. I have a better sense of what makes for good writing. In other words, the process of writing—from blank page to the final print—has made me better at editing and revising other people's work (and my own work if it's been a while since I've looked at it).

I Imagine John Warner and Tanner Greer would agree with me that if the next generation of students learn to produce essays via entering a prompt into ChatGPT and editing it, they probably will not be as good as what I am producing simply because they will not be good editors because they’ve never written from scratch.

But maybe we’re asking the wrong question here. Shouldn’t the question we’re asking be: Do student want to be good writers? And if the answer for one student is yes, then wouldn’t they follow their professor's advice and start writing from scratch rather than using ChatGPT?

The answer seems obvious. Most students are not interested in writing. And those that might be are more interested in getting a good grade as efficiently as possible. Chalk up another victory for Bryan Caplan and his theory that education is not about building human capital (ie writing skills) but about signalling (getting a degree, graduating with honors, etc.).

So as for the question of what has gone wrong, the answer is the model of college.

I want to return to a quote from Michale Lind in which he suggests breaking up colleges as others wanted to break up the Big Banks. 

"Why not just break up Big U? Why not apply the logic of antitrust to the bloated, wasteful nonprofit academic sector? Let corporations do their own vocational training, like McDonald’s with Hamburger University, or contract with free-standing trade and professional schools. Let research be done by independent science and engineering institutes, with apprenticeships for young scientists and engineers. Let gender and ethnic studies departments be left-wing nonprofits..."

A signaling device is still necessary for employers so they can tell how competent you are. But I wish that the process of obtaining that signal was not tied up in developing skills.

Imagine a world in which a student interested in journalism scores a high SAT score. The New York Times hires them in some apprentice program and pays for them to attend a writing workshop to improve their craft. Do you think that student is going to use ChatGPT to aid in their first assignment? I’d say hell no.

At that point they will actually be interested in being a better writer, a better thinker, and more open to the idea that the difficulty might be the point. And I know: that is what college is supposed to be. But, for a growing percent of students, it's not. When getting a good grade as easily as possible conflicts with building skills, most students will opt for the easy grade.

(It's also possible that none of the difficulty that Warner and Greer go through in the writing process actually matters at all and they are just doubling down on meaning.)