Thursday is Marcel Proust’s 154th birthday, so this week I’m going to talk about the very opposite of Proust: AI. Proust wrote the thousands of pages of In Search of Lost Time over years and years. Generative AI will do it in seconds… as long as you don’t plan to actually read it.
This week’s entry comes because I have a twenty-page paper due a month from today. When it comes to fiction writing, I write extremely quickly: after all, you can’t finish NaNoWriMo five times in a row without being able to make things up in a hurry. Although it’s on the high side, I’ve written 20 pages of fiction in a day before. (That’s just the rough draft, though. If you count all the time spent in revision afterward, I probably barely average a page a day.)
When it comes to non-fiction, however, I am the opposite. I like to lay out things (I actually write outlines, which is something I never do for fiction), do actual research (as in reading books and papers, not Google’s AI summary and a scan of Wikipedia), and make connections between things I’m learning and things I already know. A 20-page paper is a multi-week effort.
And here’s the temptation: in between the time I finished undergrad and the time I started grad school, there was a gap. In that gap, ChatGPT happened. I could, if I wanted, write one line (“write a 20-page research paper on the causes, decline, and consequences of the Latin American literary Boom movement”) and have 20 pages to show for it. Or if I don’t want to blatantly cheat, I could use that as “research,” which will inevitably end up being the rough draft anyway. A month’s work is done… but at what cost?
The cost, to me, is the education I’m paying for. I didn’t go to grad school to find a job; a bachelor’s degree is all the degree I need to be an airline pilot. A master’s will do nothing for me there. I’m doing this because I want to learn more about a subject I love and at the end of it, have a piece of paper that shows I know a lot about it. Taking something AI-generated and passing it off as my own doesn’t do that. It’s the difference between learning French and using a translation program. A translation AI might get you close enough, but as I’ve said before, a great translator will take you to Paris, France. A “good enough” translator will take you to Paris, Texas.
See below, where the computer doesn’t understand that “hi-hat” and “Hi, hat” aren’t the same thing. One is a part of a drum kit, the other is a way to greet the sorting apparatus at Hogwarts:
Generative AI is to the mind as GLP-1 pharmaceuticals are for the body: a way to have it all without working for any of it. Both of them have side effects that aren’t something I want to have in either my brain or my body.
Generative AI is the 21st-century version of the old paper mills that would write a term paper for you for a fee. They were marketed as “research aids” back then, too. The only difference now is that it cut the human who used to do the typing (or photocopying) out of the loop. Even if you passed the class, you learned nothing.
I was at the dentist the other day and one of the questions they asked about my medical history was whether I take a GLP-1 medication (one of the miracle weight loss drugs). My answer was, “No, I take a bicycle for that.” Generative AI is to the mind as GLP-1 pharmaceuticals are for the body: a way to have it all without working for any of it. Both of them have side effects that aren’t something I want to have in either my brain or my body.
Higher education is facing a crisis because my attitude isn’t normal. I don’t want to be a lab rat who presses a lever and gets a reward. If I lose a pound today, I want it to be because I rode 50k yesterday. If I get an A a month from now, I want it to be because I spent the next few weeks researching and writing. Higher education relies now more than ever on honor. This is a problem, and it’s one I don’t know how to fix.
I'd love to hear from you!