1
When LLMs Benchmark Themselves: Deconstructing Self-Bias in Automated Evaluation
大模型自我评估竟有系统性偏见？这篇研究拆解了自偏见的成因与影响，对AI评估可信度敲响警钟。
arXiv:2509.26600v2 Announce Type: replace-cross Abstract: As LLMs rapidly saturate existing benchmarks, automated benchmark creation using LLMs (LLM-a…