A minor sign of hope after the AI maelstrom

 AI, and the students' use of it to generate papers, consumed far too much of my brain earlier this semester. I'm teaching online, so my usual expedient of having them write in class isn't an option. 

It was wearing me out, between worrying that I was letting them get away with something and thus disadvantaging honest students or that I wasn't living up to the syllabus by checking everything. It was making me discouraged with teaching.

Turnitin wasn't helpful, nor was GPTZero, the supposedly good AI-catcher. The results could be wildly at odds with each other if you tried it twice in a row, unless something was coming up 100% AI generated. 

I called out a few students, per the syllabus. What that means: I had them talk to me. Many said it was Grammarly, which has gone heavily to AI, and said they wouldn't use it again. I am not anti-tech--eighteen years of blogging here should tell you that--but if they are not doing their own work, I can't help them make it better.

Then things started to get better. Aside from modifying the LMS discussion board settings and Perusall (no copy & paste, post your reply first before seeing others' responses--this I learned to restrict after a few students were copying from each other), I think what happened is this:

They realized that I was reading what they wrote. 

Now, I tell them this in the syllabus, but reading any syllabus, especially with all the required institutional boilerplate, is like reading the instructions for setting up a wireless router or, my favorite analogy, Beetlejuice's Guide for the Recently Deceased. 

Was it just adjusting the rubrics that made the difference? Maybe some. I discovered that having good criteria there would take care of the few AI-written posts, which naturally fell down to the C- or D level.

But I like to believe that it was that there was a real person in there, in those discussion boards, commenting and upvoting and mentioning by name the students and the specific things that they did well. They know that there is a person behind the class.

And on their papers, addressing the specifics of what they had written, suggesting other ways to develop the argument, and so on.

And in answering their emails quickly and with a sense of their real concerns. 

What I noticed is that the AI boilerplate--so far, anyway--seems to have died down, and I've mostly stopped looking for it and thinking about it.

This may, of course, just be an artifact of its being five weeks from the end of the semester, or maybe I'm missing something.

But their writing seems to be more authentic, more as it used to be, and not that MEGO AI boilerplate. 

With some of the professional organizations in the field throwing in the towel and writing guidelines not about if we will use AI but how extensively we ought to use it, I count my students' responses as a sign of hope. 

Maybe if we give them authentic feedback, as the MLA-CCCC report suggests, they will respond with authentic writing.