Expired50% confidenceFactExact time
Multiple studies in 2024 showed that LLMs exhibit a significant "self-consistency bias," tending to give overly favorable evaluations when asked to assess content they themselves generated.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026(expired)
Sources
Related Claims
Unverified「LLM-as-Judge」范式最早在2023年被系统性提出，利用大语言模型本身的评估能力对生成内容进行质量打分70% similarUnverified自我偏好偏见指裁判模型倾向偏好由同系列模型生成的输出69% similarUnverified循环论证偏差指当特征本身隐含标签信息时模型看似准确率极高实则用答案预测答案69% similarUnverified2023年多项AI研究表明，让多个LLM实例互相质疑和反驳能有效减少单一模型的幻觉问题，提升推理准确率69% similarUnverifiedA 2023 paper by OpenAI researchers found that models trained with RLHF tend to express agreement even when faced with clearly incorrect views (sycophancy bias).68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/47768API
curl https://kongchang.com/api/v1/knowledge/claims/47768MCP
get_claim(id=47768)