ICLR hits 60,000 submissions as MIT professor warns peer review is broken

CSProfKGD · x · 2026-09-22

MIT professor Phillip Torr flags ICLR's 60,000 submissions and questions what it means for AI and science: ML reviews were always noisy, good reviewers are now impossible to find, and the reciprocal review rule is "badly broken." Oral exams won't scale, and AI reviewing would just mean AI-generated papers reviewed by AIs. As ICCV'29 General Chair, he's asking for solutions.

Related event: ICLR 2027 Surpasses 60K Submissions, Sparking Reform Debate(7 posts)→

Original post →

More from Research

Research channel →