Are LLM Evaluators Really Narcissists? Sanity Checking Self-Preference Evaluations
LLM评估者真的是自恋者吗?对自我偏好评估的健全性检查
机构 * Department of Machine Learning, Carnegie Mellon University, Pittsburgh, PA, USA(卡内基梅隆大学机器学习系) ; Department of Computer Science and Engineering, University of California San Diego, La Jolla, CA, USA(加州大学圣地亚哥分校计算机科学与工程系) ; Department of Computer Science, University of Virginia, Charlottesville, VA, USA(弗吉尼亚大学计算机科学系) ; Martian Research, San Francisco, California, USA(火星研究公司) ; Apart Research, San Francisco, California, USA(Apart研究公司)
AI总结 通过比较评估者自我评价与评价其他模型时的投票分布,发现仅51%的先前结果具有统计显著性,表明自我偏好主要由评估者质量而非自恋驱动。
Comments ICML 2026 Main