A synthesis across the AI-and-teaching essays. AI removes the friction of reading, verifying, and prompting, and that friction was doing the teaching, so the work becomes deciding which frictions to protect.
Students keep calling the AI’s reading suggestions too general and Western-centric. They are right, and the reason is built into what an averaging machine is.
The struggle to write a good prompt is the struggle to know your own mind. I wonder whether that friction was the lesson all along.
We worried AI would write the student’s essay. In my research class the more interesting thing happened on the reading side, and a rented understanding is hard to tell from the real thing.
Some of my students felt bad using AI on their research. The usual move is to reassure them. I want to take the opposite seriously: maybe the discomfort is the signal worth keeping.
In my business research class, the most diligent way students used AI was to verify their own work. Confirming correctness is a judgment, and the model makes none.