Over half of NeurIPS workshop reviews look LLM-generated, sparking debate on AI peer review
anshulkundaje · x · 2026-09-28
A researcher finishing meta-reviews for a NeurIPS workshop reports a grim picture: 1 of 8 papers was largely LLM slop and more than half the reviews appeared LLM-generated, arguing that two honest lines of gut instinct beat an LLM review. Stanford professor Anshul Kundaje pushed back, saying modern LLM reviews would still be "infinitely better" than the garbage human reviews he's seen, given how rarely paper assignments match reviewer expertise. The exchange captures academia's dilemma between AI slop and chronic reviewer mismatch.
More from Fun
- Game festival vows permanent bans for any AI-generated content found in entries — AIandDesign · 2026-09-29
- First views from one of the first 26 Starlink V3 satellites deployed by Starship — PTrubey · 2026-09-29
- Two workers on quitting hollow jobs: 'hollow tenures and negative learning are real' — PAstynome · 2026-09-29
- Reddit post mocks the rise of 'AI denialism' memes — Puzzleheaded-King584 · 2026-09-28
- 13 months chasing Air Canada failed; one message from a Muse agent got the $1,000 back — armand_ruiz · 2026-09-28
- One vague hint and Codex kills a queued job: agent obedience gets a bit spooky — gowthami_s · 2026-09-28