Amy Lewis
What would I tell other professors about AI in the classroom?
For years, I graded student essays with a pad of paper next to me, and as I read, I marked down any and all grammatical errors that I saw in the papers. I didn’t mark these errors on the student papers, because I didn’t want them to think that the worth of their writing came from the correct use of commas or the lack of run-on sentences. I also wanted to avoid a revision strategy that focused their attention on such sentence-level concerns. On their papers, I wrote a paragraph or two that focused on their ideas, because then, as now, it’s the ideas and the way they express them that matter most.
If ideas matter most, why even generate a list of grammatical errors? The error list was for in-class writing workshops, where we focused on editing our final, final drafts, checking them for sentence-level errors and correcting them together. The list of errors that I generated outside of class helped me identify what to review inside the classroom. While I never once found a way to teach comma usage that didn’t fall flat, our discussion of homonyms – to, too, two; their, there, they’re; its, its – often generated a great deal of laughter. Why? I’ve no idea.
I realized that students in my classes were using AI to write their papers when there was not one grammatical error in the bunch. What I wrote on the paper next to me instead was: “Why do so many papers have this odd chirpy tone, like they’ve been written by the Church Lady on Saturday Night Live or someone who has worked in Human Resources for the past two decades?” and “Why are all the ideas so pat, so boring, so predictable?”. When a chatbot writes a paper, that paper is (unless the bot hallucinates) essentially correct, but in no way original. I realized that students had outsourced the labor involved in drafting, revising, and editing their papers, and, in doing so, they had given away their chance to think. In class, I said “I’m not going to read any more chatbot papers, because I don’t care what the bots think”. Outside of class, I thought “What should I do now?”.
I would tell other professors that teaching is dynamic profession. It changes as student populations and administrative policies change. It changes when technology changes. There’s no escaping the homework apocalypse that AI has created. Like many other teachers, I harbored an early naive belief that if a no-AI policy was clearly outlined in the course syllabus and students were asked to write personal essays, they would write such essays themselves. No, I found, many would not. Forming opinions on the assigned reading, if it happens outside of class, can be outsourced to AI. Generating questions about the assigned reading, if it happens outside of class, can be outsourced to AI. Personal reflection, if it happens outside of class, can be outsourced to AI.
However, this homework apocalypse is not the end of the world. In a time when chatbots can do so much, so accurately and so quickly, students still have a chance to reflect on and think about course materials, as long as they’re in a no-screens classroom. Reading poems, watching films, reflecting on and asking questions about the course concepts all happen in class. Exams, and especially in-class essays, also have to be done in class. In some ways, in my experience, these in-class activities have leveled the playing field. Now the students who rarely did do their homework, even when it was assigned, are participating in class. With little or no work assigned outside of class, they’re not starting the class behind students who did the homework. For many classes, there is no homework. Everyone is starting from scratch.
As a teacher living in a world the AI has changed and is changing, my next goal is to incorporate some understanding of AI literacy into the classroom. Students need to know is that it’s all about the prompt, that the clarity, specificity, and field-specific terminology of the prompt greatly influence the results. Students also need to know that AI, in its current iterations, wants to make us happy, and thus it hallucinates. A recent article in the New York Times focused on a federal judge in Mississippi, who threw out a court case after lawyers on both sides used AI to find past cases that supported their interpretation of the law. Unfortunately, none of lawyers realized that AI hallucinates. It “finds” things that don’t exist, if it thinks that’s what you want. They didn’t verify that the cases were real. And the cases were not real. With such basic AI literacy in place, I’m hoping that students will learn that AI is just one tool, in some ways an exacting and capricious tool, in a world where it is their own ideas that still matter most.
Wish me luck.
