A Simple Example
Back to my “regular programming” with a post/update/stack/whatever these things are calls that is more on brand….
Which makes this all a teachable moment for me as well: I think the lesson I’ve learned (or re-learned) from this is that the best way to prevent/discourage students from using AI to cheat is to get out in front of the issue. I’m not saying that all writing teachers ought to allow their students to use AI; in fact, as we’re approaching the end of the semester, I’m not sure if it is a good idea to encourage and sanction the use of AI in classes like first year writing. But I am sure that is is a very good idea for writing (and other kinds of ) teachers to be up-front about AI. I think when teachers do spend some time talking about what does or doesn’t work with AI, students are less likely to use it to cheat in that class— if they use it at all.
My students and I have reached the part of the semester where they are mostly working on finishing the assignments, and where I’m mostly working on reading/commenting/evaluating those assignments. So busy busy busy. Anyway, as kind of an occasional break from that work, I wrote this post in bits and pieces over the last week or two about how a particular example of AI “cheating” became a “teachable moment.”
I think there’s AI CHEATING and there’s AI “cheating,” much in the same way that there is PLAGIARISM and then there’s “plagiarism.” By PLAGIARISM, I mean the version where a student hands in a piece of writing they did not compose at all. The most obvious example is when a student pays someone else to do it, perhaps from an online paper mill. I know this happens, but I don’t think I’ve ever seen it— unless it was that good I didn’t notice.
More typically, students do this cheating themselves by copying, pasting, and slightly tweaking chunks of text from websites to piece together something kind of like the paper. This is usually easy to spot and for two reasons. First, the same Google searches students use to find stuff to cheat with also works for me to find the websites and articles they used to cheat. Second and perhaps more importantly, students only plagiarize like this when they know they’re failing and desperate, so it’s easy to spot.
The much more common kind of “plagiarism” I see is basically accidental. A lot of students— especially first year students— do not understand what needs to be cited and what does not. This is because citation is both confusing and a pain in the ass, so students sometimes do not realize they had to have a citation at all, or they just skip it and figure no one will notice. Fortunately, it’s easy to spot when students drop in a quote from an article without citation because of the writing shift: the text goes from a college freshman grappling with their prose suddenly to a polished and professional writer, often with specialized word choices and jargon. And as often as not, students do cite some of the article they’re accidentally plagiarizing, so it’s pretty easy to check.
This is a “teachable moment:” that is, one of those things that happens in a class or an assignment where it’s an opportunity to reinforce something that has already been taught. This is where I remind the student about what we already talked about: how unintentional plagiarism is still plagiarism, that this is specifically an example of why it’s important to cite your sources correctly, and so forth. This tends to click.
Similarly, there’s AI CHEATING and then there’s AI “cheating,” and I have seen examples of both in my first year writing classes this semester. The big example of extreme AI CHEATING I’ve seen so far this semester is not that interesting because it was so textbook: desperate failing student clumsily and obviously uses AI, I called the student out about it, student confesses, I gave the student the choice to fail or withdraw rather than going through the rig-a-ma-roll of getting that student expelled (oh yes, that is something I could have done). Slight tangent: if catching AI cheaters is as easy and as obvious as it seems to be, what’s the problem? Conversely, if students are using AI effectively as a tool to help their process (brainstorming, study guides, summarizing complicated texts, proofreading, etc.) and if that use of AI isn’t detectable by the teacher, well, what’s the problem with that?
The AI “cheating” example from this semester was a more interesting and teachable moment. Here’s what happened:
The first assignment in my freshman comp classes is a 2-3 page essay where students explain their initial working thesis and how they came up with it. It’s a low-stakes getting started kind of assignment I grade “complete/incomplete.” As I explain and remind students repeatedly, this is not an essay where they are trying to convince the reader to believe their thesis. Rather, this is an essay about the process of coming up with the working thesis in the first place. What I want students to write about is why they’re interested in their topic, what sorts of brainstorming activities they tried to come up with their topic, what sorts of conversations they had about this project with me and with classmates, and so forth.
This semester, the topic of research in my first year writing classes is “your career goals and AI.” I’ve also spent a lot of class time explaining why I think AI is not that useful for cheating because it just can’t do these assignments very well. But I also explained how AI might be useful as part of the process as well. For example, a lot of these students really struggle with coming up with a good and researchable topic idea/thesis, and even though most of AI’s ideas for a thesis about career goals and AI aren’t great, it does help them get beyond staring at a blank page.
I’ve given a version of this assignment for a long time, and in previous semesters and pre-AI, two or three students (out of 25) messed it up. It’s usually because the students didn’t understand the assignment, or they weren’t paying attention to/didn’t do any of the prewriting exercises we discussed in class. So they try to fake it by writing what ends up being a really short research paper without any research. I gave these students a do-over, and that usually was enough to get them back on track.
This semester, I had closer to half of the students in my two sections mess this up. I’m sure some of these students just didn’t get the assignment/didn’t do the prewriting activities, but what I think happened more is a lot of students got a little lazy and hypnotized by the smooth, mansplaining prose of AI. So instead of remembering what the assignment was about, they just took what the AI was feeding it about their working thesis ideas and tweaked that a bit.
The teachable moment? I met with the students who messed this up, reminded them what the assignment was actually supposed to be, and I pointed out that this was exactly the kind of thing that AI cannot do: it can’t help you write about what you think. At least not yet.
This was a couple weeks ago, and for most of my students, I think it clicked. I still have a number of students who are struggling and unlikely to pass for all kinds of reasons, but that’s typical for freshman comp. Some students (particularly the ones on the way to failing) are still trying to use AI for cheating, but for the most part, I think students have learned the lesson.
I ask students to include an “AI Use Statement” where they describe how they used AI, or to say explicitly that they didn’t use any AI. This is a brand-new thing for both them as students and me as a teacher, so they sometimes forget or they don’t explain their AI use as clearly as I wanted. And I am sure some students are fibbing a little about how much AI they used. But for the most part, what students are telling me is they aren’t using AI to write at all, or they’re using Grammarly for proofreading (which I think counts as AI), they are using an AI for some ideas about a particular paragraph, and/or getting started or some other brainstorming kind of suggestion.
Which makes this all a teachable moment for me as well: I think the lesson I’ve learned (or re-learned) from this is that the best way to prevent/discourage students from using AI to cheat is to get out in front of the issue. I’m not saying that all writing teachers ought to allow their students to use AI; in fact, as we’re approaching the end of the semester, I’m not sure if it is a good idea to encourage and sanction the use of AI in classes like first year writing. But I am sure that is is a very good idea for writing (and other kinds of ) teachers to be up-front about AI. I think when teachers do spend some time talking about what does or doesn’t work with AI, students are less likely to use it to cheat in that class— if they use it at all.