TMLR Editor Quizzed 10 Authors About Their Own Papers — Most Couldn't Answer

机器之心 · wechat · 2026-09-26

TMLR co-editor-in-chief Nihar Shah interviewed authors of 10 desk-rejection candidates about their own papers; 3 couldn't answer basic questions, post-hoc replies from two were flagged 100% AI-generated. TMLR submissions tripled (single-author up 13x), desk rejections hit 53%, and the journal rolled out author quotas, AI reliability reviews, and clarity requirements — part of a wider crisis including NeurIPS' 28% AI-flagged track and arXiv's one-year bans.

Original post →

More from AGI Musings

AGI Musings channel →