Research
Twice-Told Critiques: Claim-Level Homogenization of Peer Review in the LLM Era
The study measures overlap among critiques in conference and journal peer reviews and reports rising overlap at ICLR, NeurIPS, ICML, F1000 and PLOS. At ICLR, measured overlap rose from 12.0% in 2026 to 21.3% in 2027, while a third review added 5.46 new problems in 2027 versus 6.12 in 2026. Reviews flagged by the MELD LLM-text detector overlapped more than unflagged reviews within the same paper, suggesting falling critique diversity.
peer reviewLLM-generated text detectionscientific publishing