Current limits of AI peer review in judging novelty
A user sharing work by Zhang and colleagues says it argues that judging novelty remains beyond even otherwise capable AI models.
TLDR
The paper’s arXiv summary says large language models are increasingly used in peer review at major AI conferences, but novelty remains a persistent weak point. A user sharing the work by Zhang and colleagues welcomes it as a concrete examination of AI peer review, describing its argument as a limitation of current models: they cannot yet judge novelty.
Combined views
1.9K
1 Source, first seen 16d ago