LLM Peer Reviewers Fixate on Appendices, Unlike Humans, Study Finds

CincyNLP · hf · 2026-11-16

A study by CincyNLP finds that large language models used for peer review draw half or more of their identified weaknesses from paper appendices, while human reviewers derive only 8.4% of critiques from appendices and focus instead on presentation and core ideas — revealing a systematic bias in LLM-based reviewing.

Original post →

More from Models

Models channel →