Author flags wrong ICLR review, reviewer drops score to 1 and pastes a fully AI-generated critique

menhguin · x · 2026-11-08

An ICLR 2027 author discovered a reviewer had reviewed the wrong paper; after being politely told, the reviewer lowered the score to 1, raised confidence to 4, and pasted in a 100% AI-generated review.

The quote-tweet quips that such errors make one grateful it was "at least human error" — with the follow-up that the AI review was probably just bad copy-paste. A vivid case of peer-review breakdown and AI-slop reviewing in academia.

Original post →

More from Fun

Fun channel →