LLM agent groups overstate consensus in reasoning tasks, study finds

A new study published on arXiv suggests that large language model (LLM) agent groups may overstate consensus when simulating human deliberation on reasoning tasks. When replaying human groups on the Wason task, LLM agents consistently showed higher consensus rates, even when accounting for differences in participation and operationalization. This simulated consensus did not accurately track collective accuracy, with agent groups often agreeing on incorrect answers, indicating a potential bias in their estimation of human group outcomes. AI

影响 Highlights potential biases in LLM simulations of human group reasoning, suggesting caution when interpreting AI-generated consensus.

排序理由 The cluster contains an academic paper published on arXiv detailing research findings. [lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• ScienceCast
• Wason

• 论文
• 安全

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →