研究称 Claude Opus 4.8 生成的论文评审已全面优于人类专家
精选理由
Anthropic 的 Opus 4.8 写的论文评审，在所有指标上赢过了期刊公认的高质量人类专家评审，同行评审要变天了。
一项对比研究显示，Claude Opus 4.8 生成的论文评审在研究者构造的所有衡量指标上都超过人类专家的评审。这些人类评审来自薪酬丰厚的专家，且曾被期刊编辑和作者评价为质量格外高。该结果指向 AI 在学术出版评审环节的实际可用性。
原文 · Teortaxes
«AI-generated article reviews are now better, by every measure we could construct, than reviews written by well-compensated human experts and judged to be of unusually high quality by journal editors and authors.»
By AI they mean Opus 4.8 btw posthuman science [publishing] comes https://t.co/U3WdMRw59U
- Ethan Mollick04:10原文
