AI reviewers favor confidently written science papers, study finds

A German study found that LLM peer reviewers give higher scores to scientific papers written in confident language, even when the underlying research is identical. In 21% of cases, Claude Sonnet 4.6 recommended accepting a confident version while rejecting a cautious rewrite of the same paper.

Unite.AI

Open source

AI reviewers favor confidently written science papers, study finds

504 ICLR 2026 papers were rewritten in confident and cautious styles using Claude Opus 5
Six models scored them, including GPT-5.5, Claude Sonnet 4.6 and Gemini 3.5 Flash
Claude Sonnet 4.6 favored confident versions seven times more often
In 21% of cases the verdict flipped from reject to accept on wording alone

#Claude#GPT-5.5#Gemini#ICLR

AI reviewers favor confidently written science papers, study finds

AI Models Still Fail to Properly Correct Users, Study Finds

Study: Claude Sonnet 4.6 and GPT-5.5 Show Gender Bias in Moral Judgments, DeepSeek V4-Flash Doesn't

Study: 2,500 telltale phrases reveal text written by Claude Opus 5.5

Microsoft pushes to build AI confidence in the UK