Now that the 2023-24 school year is long over and my wife and I are (mostly) done moving into our new house, it’s time to start thinking again about AI for teaching in the fall and for some scholarly things beyond. I’ve been mostly ignoring these things for the last couple of months, but even in that short time, it feels like things have changed. AI tech is getting quickly integrated into everything you can imagine, and it feels to me like the AI faculty freakout factor is on the rise once again.
This is just a gut feeling– like I said, I’ve been out of the loop and it’s not like I’ve done any research on this. But the current moment reminds me a bit of late 2022/early 2023 when ChatGPT first appeared. By the time I did a talk about AI at Hope College in late April 2023 and also again a talk/workshop about AI (over Zoom) at Washtenaw Community College in October 2023, teachers had settled down a bit. Yes, faculty were still worried about cheating and the other implications, but I think most of the folks who attended these events had already learned more about AI and had started to figure out how to both use it as a tool to help teaching. They also realized they needed to make some changes to assignments because of AI tools.
But now the freakout is back. Perhaps it’s because more faculty are starting to realize that “this whole AI thing” is something they’re going to have to deal with after all. And as far as I can tell, a lot of the freaked out faculty are in the humanities in general/in English in particular. I suppose this is because we teach a lot of general education classes and classes that involve a lot writing and reading. But I also think that the reason why the freakout is high in fields like English is because a lot of my department and discipline colleagues describe themselves as being “not really into technology.”
The primary freakout then and now– at least among faculty in the humanities (I assume STEM faculty have different freakout issues)– is that AI makes it impossible to teach writing in college and in high school because it is too easy for students to have ChatGPT (or whatever other AI) to do the work for them. I wrote a post in response to these articles back in December 2022, but there were dozens of freakout articles like these two. These articles almost always assume that AI has uniquely enabled students to cheat on assignments (as if paper mills and copy and pasting from “the internet” hadn’t existed for decades), and that given the chance, students will always cheat. So the only possible solution is to fight AI with things like detection software or returning to handwritten exams.
It’s deja vu all over again.
Consider, for example, Lisa Lieberman’s June 2024 Chronicle of Higher Education article “AI and the Death of Student Writing.” Lieberman, who teaches community college English and composition courses “in California’s Central Valley,” has seen an alarming uptick in students using AI to write their papers. She gives an example of a student’s essay about The Shining that included the sentence “A complex depiction of Jack’s development from a struggling family guy to a vessel of lunacy and malevolence is made possible by Stanley Kubrick’s brilliant direction.” Lieberman writes “I called the student in and asked him to write a sentence with the word ‘depiction.’ He admitted he didn’t know what ‘depiction’ meant, much less how to spell it, much less how to use it in a sentence. He confessed he hadn’t written a single word of the essay.” (For what it’s worth, I would have asked this student about “malevolence”).
Then she moves on to discussing a student writing her essay with the now AI-fueled version of Grammarly. Lieberman “discovered it’s a multilayered computer program that does everything from simple spelling and grammatical corrections to rewriting entire sentences, adjusting tone and fluency.” She estimated that at least of a third of her students were consistently using AI: “Once they believed they could turn in AI assignments undetected, they got bolder … and used AI for every single assignment.”
It’s all just so wrong, Lieberman laments, in part because of how her students are just cheating themselves by using AI. Here’s a long quote from the end of the article:
I remember my days at Berkeley, where, as an English major, I’d take my copy of Wallace Stevens’s The Palm at the End of the Mind, or Chaucer’s “The Wife of Bath’s Tale,” and pick a nice, sunny spot on campus on a grassy knoll underneath a tree, lay out my blanket, and spend the afternoon reading and scribbling notes in my books. It was just me and my books and my thoughts. There was nothing better.
As I lay there reading the writer’s words, they came to life — as if the author were whispering in my ear. And when I scribbled my notes, and wrote my essays, I was talking back to the author. It was a special and deep relationship — between reader and writer. It felt like magic.
This is the kind of magic so many college students will never feel. They’ll never feel the sun on their faces as they lie in the grass, reading words from writers hundreds of years ago. They won’t know the excitement and joy of truly interacting with texts one-on-one and coming up with new ideas all by themselves, without the aid of a computer. They will have no idea what they’re missing.
I understand the anxiety that Lieberman is expressing, and I completely agree that AI technology is forcing us to change how we teach college classes– and, in particular, classes where students are expected to read and to write about that reading.
However:
- Students have been cheating in school for as long as there has been school. AI make it easier (and more fun!) to cheat, but none of this is new. So any educator who thinks that students have only now started to cheat on the things they assign only because of AI are kidding themselves.
- In my experience, the vast majority of students do not want to cheat this much. Oh sure, they might cheat by poorly borrowing a quote from a website, or looking over someone’s shoulder to get a quiz answer on a multiple choice test. But in my view, these are misdemeanor offenses at best. Also, when students do not cite sources properly (and this is as true for the MA students I work with as it is with the first year writing students), it’s because they don’t know how. In other words, a lot of plagiarism is a teachable moment.
- Also in my experience, students who do blatantly cheat by downloading from a papermill or prompting an AI to do the whole assignment are a) already failing and desperate, and b) not exactly “criminal masterminds.” Every freakout narrative I’ve read– including Lieberman’s– includes a “scene” where the instructor confronts the student with the obvious AI cheating. So to me, if it’s this easy to catch students who cheat using AI, what’s the problem? Just punish these students and be done with it.
- The fundamentals of teaching writing as a process– the mantra of writing studies for the last 50+ years– are still the same and the best way to discourage students from cheating with AI or anything else. Don’t merely assign writing– teach it. Make students show their work through drafts. Use a series of shorter assignments that build to a larger and more complex writing project. In a research-oriented writing class (like first year composition, for example), require students to create an annotated bibliography of all of their sources. Have peer review as a required part of the process. Etc., etc., etc. None of this is foolproof and for all I know, Lieberman is already doing this. But besides actually helping students to become better writers, teaching (rather than just assigning) writing like this makes cheating as much work as just doing the assignments.
- I think the best way to dissuade students from using AI to cheat is to explain to them why this is a bad idea. Last year, I had a discussion at the beginning of all of my classes on the basics of AI and why it might be useful for some things (see my next bullet), and why it is not useful for cheating, and that’s especially true in classes that involve research and where writing is taught as a process (see my previous bullet). I think by making it clear from the beginning that yes, I too knew about AI and here’s why cheating with it isn’t a good idea, fewer of them were tempted to try that in my classes.
- I don’t think there’s anything wrong with Grammarly. At EMU, I will often get letters of accommodation from the disability office about students enrolled in my classes that tell me how I am supposed to “accommodate” the student. That usually means more time to take exams or more flexibility for deadlines, but often, these letters say I should allow the student to use Grammarly.
My philosophy on this has always been that it is a good idea for students to seek help with their writing assignments from outside of the class–help that assists, not that does the work for the student. I always encourage students– especially the ones who are struggling– to get help from a writing center consultant/tutor, a trusted friend or parent, and so forth. I think Grammarly– when used properly– falls into that category. I don’t think asking Grammarly to write the whole thing counts as “proper use.” I want students to proofread what they wrote to make sure that the mechanics of their writing are as clear and “correct” as possible, and if Grammarly or an AI or another electronic tool can help with that, I’m all for it.
I think the objection that Lieberman has with Grammarly is it makes writing mechanically correct prose too easy, and the only way for students to learn this stuff is to make them do it “by hand.” As someone who relies heavily on a calculator for anything beyond basic arithmetic and also as someone who relies on Google Doc’s spell checking and grammar checking features, I do not understand this mindset. Since she’s teaching in a community college setting, I suppose Liberman might be working more with “basic writing” students. I could see more of an argument for getting students to master the basics before relying on Grammarly. But for me and even in classes like first year writing, I want to focus mostly on the arguments my students are making and how they are using evidence to support their points. So if a student gets some help with the mechanics from some combination of a writing center consultant and an application like Grammarly, then I can focus more exclusively on the interesting parts.
Where Lieberman and I might agree though is if a student doesn’t have basic competency with writing mechanics, then Grammarly is not going to solve the problem. It’s a lot like the mistakes students still make with there/their/they’re even if they take the time to spell check everything. And again, that’s why it is is so easy to detect AI cheating: the vast majority of students I have had who have tried to cheat with AI have done it poorly.
- Finally, about students missing “the magic” of reading and writing, especially while doing something
clichédidealistic like laying on a blanket on the campus lawn and under an impressive oak. I get it, and that’s part of why I went into this line of work myself. But this is the classic mistake so many teachers make: just because the teacher believes reading and writing are magical doesn’t mean your students will. In fact, in required gen ed classes like first year writing or intro to literature, many (sometimes most) of the students in those classes really do not want to take those courses at all. I can assign students to read a book or essay that I think is great or I can encourage students to keep writing on their own and for not just school, and sometimes, I do have students who do discover “the magic,” so to speak. But honestly, if the majority of my first year writing students at the end of the semester come away thinking that the experience did not “totally suck,” I’m happy.
So no, this is not the end fo student wri
One Reply to “No, Student Writing Is Not Dead (or how AI faculty freakout is back)”