It's Not about Cheating

What the general public gets wrong about AI in the classroom

Recent conversation with a contemporary (a man who worked in sales all his life, and whose grandchildren attend a Christian school):

Him: So what do you think about AI? How will your public schools deal with the fact that AI is going to control all jobs in the future?

Me: AI will certainly have an impact on the job market, but I don’t think the future of work is written in stone. As with all technologies, experience will tell us whether AI is actually useful in enhancing learning in any way. Lots of things that sound good in education turn out to be oversold or hype. Or even counterproductive.

Him: But isn’t AI going to make it impossible to tell who’s cheating? That’s what I’d be worried about if I was a teacher.

Me: What do you mean by cheating?

Him: Well, kids will get AI to write their papers and do their assignments. And teachers won’t know who wrote the paper and will be forced to give it a good grade. And if everyone gets good grades, there will be grade inflation, so it will be hard to pick out the really smart students for the top colleges.

Me: It’s not about cheating. It’s about actual learning. Students learn by doing the work, including making mistakes—whether that work is putting two blocks with three blocks to make five blocks, or testing pond water samples, producing an original haiku in class–or writing a research paper. When people talk about AI and cheating, they’re usually thinking about writing assignments—but there are many more paths to learning, K-12, than writing a paper or answering questions on a worksheet. Besides, teachers who know their students well, and have seen their skills in action, will understand how an AI-constructed response would compare to an actual response.

Him: (dubious) I suppose sharp teachers can catch them that way. Besides, you’ll have more time to ferret out cheaters when AI starts grading student work and writing your lesson plans.

Me: Only someone who knows the students and knows the usual flow of content and skills at that level can write useful lesson plans. And assessing student work is how teachers observe what their students have learned, and what they need next. I personally don’t see AI as being particularly useful in developing instructional materials, either. It certainly can’t develop relationships with kids or inspire them.

Him: Of course, this would all be different for you, as a band director—AI will change everything for regular teachers but maybe not for you. If band even exists as a class any more.

———–

Sigh. This conversation actually happened. And the man I was talking to was not an idiot. He had some magazine-article background knowledge about AI, saw its impact as inevitable and teachers as unfortunately unionized Luddites, unwilling to adapt to a rapidly changing world.

He was also right about musical performing groups—as a K-12 musical specialist, I have been having these conversations about electronic alternatives to learning to play an instrument or sing for three decades. Who would want to go to the trouble, a well-meaning friend who teaches English asked me, to learn to play the bassoon? Or even worry about singing in tune, now that auto-tune is available to fix hot musical stars’ vocal uncertainties?Why not grab a bunch of keyboards and software? Isn’t that all the instruction musicians need to, you know, put out musical content?

The great danger of using the range of AI products in the classroom has nothing to do with cheating, per se. Fact is, students have been cheating—in the ways we usually perceive as academic cheating—forever.

From writing dates on a shirt cuff to paying someone to take your SATs, cheating is deeply embedded in academic practice. If there is a potentially positive outcome here, it might be disconnecting old ideas about plagiarism and cheating. Instead, we might be teaching our students to assess information they are presented with, comparing it to different analyses, perhaps rooting out alternative facts that aren’t really factual.

Fact is: plagiarism is ill-defined, in an era when students have access to the Library of Congress in their raggedy jeans pockets: “Anybody who embarks on a study of plagiarism hoping for bright lines is in for a foggy shock. One of the pleasing facets of plagiarism is that it doesn’t exist—not in the eyes of the law, that is, and especially not if those eyes are American. There is intellectual-property law, and a law that prohibits the trafficking of counterfeit goods. There are laws against copyright infringement. If plagiarists are sent to prison, however, it will not be because they have filched a slice of poetry, or half a juicy ballad, and passed it off as their own. Plagiarism is not a crime. It is a sin.’”

Here’s another fact: Large language models that support the kinds of AI K-12 teachers and students are being urged to adopt are constructed of plagiarized, if you will, content. Speaking of cheating.

But it’s the original point that matters most here: AI in its various platforms robs students of doing the actual work of learning: absorbing, comprehending, analyzing, synthesizing and so on. I would like to think that this is the reason that states and school districts are banning the use of cellphones in the classroom—to prevent students from believing that graded products represent actual learning.

I would also assert that learn-by-doing classes that require groups of learners (like band and choir, debate, drama and so many others) reward students for all the right habits: working together, interdependence, ongoing skill building toward a clear goal, aesthetic pleasure. Creativity, the antithesis of AI use.

Philosophy professor Kate Manne wrote a terrific piece about preventing her university students from using AI, and how it all worked out: “I feel strongly, as I explained, that their AI use will prevent me from doing my job in helping them to grow as thinkers and writers.” Spoiler alert: students produced such superior work and thinking that she cancelled the final exam. Read the piece. It’s solid evidence.

Pushback against AI is not and never has been about cheating. It’s about genuine learning.