The Problem is Not the AI

The other day, I heard the opening of this episode of the NPR call-in show 1A, “Know It All: ChatGPT In the Classroom.” It opened with this recorded comment from a listener named Kate:

“I teach freshman English at a local university, and three of my students turned in chatbot papers written this past week. I spent my entire weekend trying to confirm they were chatbot written, then trying to figure out how to confront them, to turn them in as plagiarist, because that is what they are, and how I’m going penalize their grade. This is not pleasant, and this is not a good temptation. These young men’s academic careers now hang in the balance because now they’ve been caught cheating.”

Now, I didn’t listen to the show for long beyond this opener (I was driving around running errands), and based on what’s available on the website, the discussion  also included information about incorporating ChatGPT into teaching. Also, I don’t want to be too hard on poor Kate; she’s obviously really flustered and I am guessing there were a lot of teachers listening to Kate’s story who could very personally relate.

But look, the problem is not the AI.

Perhaps Kate was teaching a literature class and not a composition and rhetoric class, but let’s assume whatever “freshman English” class she was teaching involved a lot of writing assignments. As I mentioned in the last post I had about AI and teaching with GPT-3 back in December, there is a difference between teaching writing and assigning writing. This is especially important in classes where the goal is to help students become better at the kind of writing skills they’ll need in other classes and “in life” in general.

Teaching writing means a series of assignments that build on each other, that involve brainstorming and prewriting activities, and that involve activities like peer reviews, discussions of revision, reflection from students on the process, and so forth. I require students in my first year comp/rhet classes to “show their work” through drafts that is in a way they similar to how they’d be expected to in an Algebra or Calculus course. It’s not just the final answer that counts. In contrast, assigning writing is when teachers give an assignment (often a quite formulaic one, like write a 5 paragraph essay about ‘x’) with no opportunities to talk about getting started, no consideration of audience or purpose, no interaction with the other students who are trying to do the same assignment, and no opportunity to revise or reflect.

While obviously more time-consuming and labor-intensive, teaching writing has two enormous advantages over only assigning writing. First, we know it “works” in that this approach improves student writing– or at least we know it works better than only assigning writing and hoping for the best. We know this because people in my field have been studying this for decades, despite the fact that there are still a lot of people just assigning writing, like Kate. Second, teaching writing makes it extremely difficult to cheat in the way Kate’s students have cheated– or maybe cheated. When I talk to my students about cheating and plagiarism, I always ask “why do you think I don’t worry much about you doing that in this class?” Their answer typically is “because we have to turn in all this other stuff too” and “because it would be too much work,” though I also like to believe that because of the way the assignments are structured, students become interested in their own writing in a way that makes cheating seem silly.

Let me just note that what I’m describing has been the conventional wisdom among specialists in composition and rhetoric for at least the last 30 (and probably more like 50) years. None of this is even remotely controversial in the field, nor is any of this “new.”

But back to Kate: certain that these three students turned in “chatbot papers,” she spent the “entire weekend” working to prove these students committed the crime of plagiarism and they deserve to be punished. She thinks this is a remarkably serious offense– their “academic careers now hang in the balance”– but I don’t think she’s going through all this because of some sort of abstract and academic ideal. No, this is personal. In her mind, these students did this to her and she’s going to punish them. This is beyond a sense of justice. She’s doing this to get even.

I get that feeling, that sense that her students betrayed her. But there’s no point in making teaching about “getting even” or “winning” because as the teacher, you create the game and the rules, you are the best player and the referee, and you always win. Getting even with students is like getting even with a toddler.

Anyway, let’s just assume for a moment that Kate’s suspicions are correct and these three students handed in essays created entirely by ChatGPT. First off, anyone who teaches classes like “Freshman English” should not need an entire weekend or any special software to figure out if these essays were written by an AI. Human writers– at all levels, but especially comparatively inexperienced human writers– do not compose the kind of uniform, grammatically correct, and robotically plodding prose generated by ChatGPT. Every time I see an article with a passage of text that asks “was this written by a robot or a student,” I always guess right– well, almost always I guess right.

Second,  if Kate did spend her weekend trying to find “the original” source ChatGPT used to create these essays, she certainly came up empty handed. That was the old school way of catching plagiarism cheats: you look for the original source the student plagiarized and confront the student with it, court room drama style. But ChatGPT (and other AI tools) do not “copy” from other sources; rather, the AI creates original text every time. That’s why there have been several different articles crediting an AI as a “co-author.”

Instead of wasting a weekend, what Kate should have done is called each of these students into her office or taken them aside one by one in a conference and asked them about their essays. If the students cheated,  they would not be able to answer basic questions about what they handed in, and 99 times out of 100, the confronted cheating student will confess.

Because here’s the thing: despite all the alarm out there that all students are cheating constantly, my experience has been the vast majority do not cheat like this, and they don’t want to cheat like this. Oh sure, students will sometimes “cut corners” by looking over to someone else’s answers on an exam, or maybe by adding a paragraph or two from something without citing it. But in my experience, the kind of over-the-top sort of cheating Kate is worried about is extremely rare. Most students want to do the right thing by doing the work, trying to learn something, and by trying their best– plus students don’t want to get in trouble from cheating either.

Further, the kinds of students who do try to blatantly plagiarize are not “criminal masterminds.” Far from it. Rather, students blatantly plagiarize when they are failing and desperate, and they are certainly not thinking of their “academic careers.” (And as a tangent: seems to me Kate might be overestimating the importance of her “Freshman English” class a smidge).

But here’s the other issue: what if Kate actually talked to these students, and what if it turned out they either did not realize using ChatGPT was cheating, and/or they used ChatGPT in a way that wasn’t significantly different from getting some help from the writing center or a friend? What do you do then? Because– and again, I wrote about this in December— when I asked students to use GPT-3 (OpenAI’s software before ChatGPT) to write an essay and to then reflect on that process, a lot of them described the software as being a brainstorming tool, sort of like a “coach,” and not a lot different from getting help from others in peer review or from a visit to the writing center.

So like I said, I don’t want to be too hard on Kate. I know that there are a lot of teachers who are similarly freaked out about students using AI to cheat, and I’m not trying to suggest that there is nothing to worry about either. I think a lot of what is being predicted as the “next big thing” with AI is either a lot further off in the future than we might think, or it is in the same category as other famous “just around the corner” technologies like flying cars. But no question that this technology is going to continue to improve, and there’s also no question that it’s not going away. So for the Kates out there: instead of spending your weekend on the impossible task of proving that those students cheated, why not spend a little of that time playing around with ChatGPT and seeing what you find out?