Karpathy says AI‑homework crackdown failed, urges in‑class grading shift

Andrej Karpathy, the former Tesla AI chief, has taken a blunt stance on the recent wave of school policies aimed at catching AI‑generated assignments. He argues that the “war on AI homework” has not delivered the results administrators hoped for, and that the effort to police every submission is both costly and ineffective. While the intent was to preserve academic integrity, Karpathy points out that students quickly learn to circumvent detection tools, leaving teachers with a false sense of security.

He suggests that the focus should move away from after‑the‑fact checks and toward a model where assessment happens under direct supervision. In his view, this shift would make it harder for learners to rely on language models during evaluation and would reinforce the habit of solving problems unaided. The implication is clear: if schools want to keep learning authentic, they need to redesign how they grade.

As a result, Karpathy believes the majority of grading has to shift to in‑class work, where teachers can physically monitor students. This change keeps students motivated to learn how to solve problems without AI, since they know they will be evaluated without it later. Karpathy stresses that he's n