Last week I posted some thoughts on AI in the classroom. It was an uncontroversial post. Far less so than earlier ones where I got into actual student use of AI. This one was simpler: AI is good at finding patterns in data, so use that to quickly spot where students are struggling and where they could use help.

The analogy I like: physicians are using AI to read mammograms, catching cancers earlier and cutting down on false positives. Critically, AI isn't diagnosing or treating patients, the physician still does that. AI is simply a tool the physician uses to get better outcomes.

That analogy isn't far off from education. I suggested educators could use AI to review exams, quizzes, papers, and so on, to spot patterns and catch trouble early. And just like the radiology case, AI doesn't decide the course of action. It doesn't implement remediation. It doesn't interact with the student at all. The teacher does that. AI just helps the teacher do what they've always wanted to do: provide individualized instruction and assistance, efficiently and at scale.

As I said, uncontroversial. Or so I thought.

Apparently not. I riled up a whole lot of folks, and their responses were generally of the "you suck" and "Please go away" variety. One told me to "spend time as an educator in a K-12 classroom." I have, actually. On and off, for four decades. I just hadn't previously been asked to leave one for suggesting AI could help teachers spot struggling students sooner.

My favorite: "Students who won't do work or engage with a teacher won't do it for a computer." That would be a fair point, if the post had suggested replacing teachers with computers. It didn't. Go back and read it: the whole argument is that the teacher does the engaging. AI never talks to the student at all.

Based on the timestamps and referrer logs, that tracks: most of these reactions came from people who hadn't actually read the post. A headline and an excerpt were all it took to trigger them. And that comment about students and computers is the clearest example, it's not disagreeing with what I wrote, it's disagreeing with something I never said.

That's exactly what I mean when I talk about the AI backlash casting too broad a net. I called it "banned by name, guilt by association": react to the word AI, not to the actual proposal. Which means I should probably thank the people leaving these comments: the kneejerk responses didn't just illustrate that point, they proved it.

And that's the real shame here. AI, used wisely and cautiously, really can help our students. But that conversation can't happen if the reaction arrives before the reading does.