Megadose The AI curve, compressed.

AI Reviews Are Biased Towards the Appendix

· ArXiv · AI/CL/LG ·
GPT and Claude drew about half or more of their review weaknesses from appendices; human reviewers drew 8.4%.

The study examined 2,000 papers from ACL, EMNLP, COLM and ICLR. The pattern held across API, coding-agent and web access methods. Models criticized fine-grained technical details deep in appendices that were not central to the papers’ contributions. Human appendix criticisms primarily concerned presentation and content related to a paper’s core idea.

ArXiv · AI/CL/LG's note

score 5

Categories: Research