Unverified50% confidenceOpinionExact time
LLM在生成摘要时天然倾向于反映评论中的多数意见，混淆了信息频率与信息重要性
1
Sources
50%
Confidence
Long-term
Relevance
7/9/2026
First Seen
Sources
Tripadvisor AI摘要为问题酒店"洗白"？旅行者必看的安全隐患
hackernewshackernews7/5/2026
Related Claims
Unverified普林斯顿大学等机构的早期研究表明，内容的权威性信号、结构化表达、以及被高质量来源引用的频率会影响LLM在生成答案时对品牌的提及概率72% similarUnverified同一LLM在同一次推理中同时扮演创作者和批评者容易陷入自我确认偏误，倾向于认可自己生成的内容71% similarUnverifiedLLM evaluation is more objective than human evaluation, as it can score candidates consistently within a unified evaluation framework, avoiding the unfairness caused by cognitive biases such as fatigue, mood, and the primacy effect in human interviewers.68% similarUnverifiedLLM存在自信的流畅性（Confident Fluency）倾向，倾向于给出听起来正确的答案而非承认不确定性67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/330721API
curl https://kongchang.com/api/v1/knowledge/claims/330721MCP
get_claim(id=330721)