ICLR debate: cap author rebuttals — and reviewers' 32 AI-generated weaknesses too?

TuhinChakr · x · 2026-11-19

A debate over ICLR review norms: weilunchao argues the conference should cap author rebuttals, since a paper needing an extremely long rebuttal 'just means it's not ready,' and lengthy rebuttals tempt reviewers to use 'tools' to cope. TuhinChakr raises the symmetric problem: should reviewers also be limited from dumping 32 AI-generated weaknesses? The thread highlights how rebuttal arms races and AI-generated reviews are feeding each other in peer review.

Original post →

More from Research

Research channel →