Tuesday, November 5, 2024

The Tells of AI Writing

Universities are working at the speed of, uh, universities, to help teachers counter students submitting work completed by artificial intelligence.

Students, of course -- at least those looking for shortcuts -- are on AI's bleeding edge, using all sorts of AI tricks to make their "work" easier.

But here's the thing: Detecting artificial intelligence writing is pretty easy, even without AI detectors.

I'm not going to reveal my secrets here. First because there are likely developers on the bleeding edge of AI who are already aware of AI's weaknesses, but also because in case they're not, I don't want to aid and abet. But AI writing is obvious on its face, and is mediocre writing at best.

Part of me wants to continue calling out students who use artificial intelligence to write their essays.

A growing part of me, however, is leaning toward just grading the AI writing and giving students the grades they've earned with their "writing."

Not necessarily Fs, of course, because AI isn't that bad. But it's not churning out A level work. Or even B level work. And in many cases, barely C level work. So use AI in my classes, and you're going to get the grade you deserve, whether I "detect" your use of artificial intelligence or not.

Agreed, that's not what I'm supposed to do. I'm supposed to fail those students, report them to the university and all. But I'm not sure that works to deter the behavior, even if they get expelled from the university. But getting mediocre grades for AI writing -- and consistently mediocre grades, time after time -- might make them think about how good AI is.

And maybe I'm fooling myself. Could be. But it's more entertaining that way.

No comments: