BY JONATHAN REES
I wrote the first draft of “The AI Nuisance,” my contribution to the latest issue of Academe on artificial intelligence, back in November. The AI news kept coming at such a fast pace that I felt the need to revise it again in mid-January, just before it was due at the end of that month. My great worry was that something I read between then and the date of its eventual publication would render my argument obsolete. That didn’t happen.
If I had to summarize the argument of my article, it would be that even though computer programs can now write proper English sentences, a lot of proper English sentences randomly strung together don’t constitute a college essay that deserves a passing grade. AI can’t write an argument because it can’t think. Therefore, those of us who receive AI-composed essays from our students should just give them the extremely bad grades that they all deserve.
Nothing I read between January and May changed my mind about this argument about AI writing. However, there was one set of stories I saw that made me think that this argument might be expandable beyond the large language models that have stressed out so many professors across the country. On April 27, 404 Media published an article called “University Professors Disturbed to Find Their Lectures Chopped Up and Turned Into AI Slop.”
A new program called “Atomic,” created by Arizona State University, grabbed parts of lectures from the learning management system there and created new lessons out of them. They were made from “out-of-context, extremely short clips some cases,” and nobody consulted the faculty who created the original material. “Furthermore,” explained reporter Samantha Cole, “testing I and others did on Atomic showed academically weak and even inaccurate content.”
Remind you of anything?
Here’s how ASU President Michael Crow described this program, which is still in beta testing, to the Arizona Board of Regents:
“Imagine that we have thousands and thousands and thousands of courses. And you can break these courses down into tens of thousands or hundreds of thousands of sub-component parts.
“Then you build a program in which you can ask the computer, ‘I want to learn about this.’ And then it takes some component of all these different things and then organizes what you need to learn.”
Crow added, “I don’t know if this is going to work or not.”
It won’t.
Computers can’t think. Therefore, they can’t carry an argument from the beginning of an essay until the end, and they can’t do it with video either. You can’t replace professors with AI slop and expect students to pay for it, because the education that AI provides isn’t a real education at all.
Undergraduates cheating on their writing assignments may get most of the media attention, but the real AI problem is university administrators who think they know more about teaching than their professors do.
Jonathan Rees is professor of history at Colorado State University–Pueblo. He is copresident of the CSU Pueblo AAUP, a former copresident of the Colorado AAUP conference, and a former two-term member of the AAUP’s national Council. His email address is jonathan.rees@csupueblo.edu.
