conformity-bias

Tag

Cards List
#conformity-bias

Language Models Show Conformity Bias in RAG: When the Majority Endorses a Poisoned Claim

arXiv cs.CL ↗ · 2d ago Cached

该论文研究了RAG系统中的从众偏差:当检索到的多数文档是互相一致的投毒文档时,Gemma-3-12b检测中毒声明的能力从25%显著下降至10%,表明语言模型会随附多数证据而忽视事实核查。

0 favorites 0 likes
← Back to home

Submit Feedback