Unverified80% confidenceFactExact time
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
1
Sources
80%
Confidence
Long-term
Relevance
5/23/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedLLM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72% similarUnverifiedHuman annotators exhibit a cognitive bias toward more information being better when evaluating answers, causing reward models to implicitly link length with quality, which is one of the causes of model verbosity.66% similarUnverified2023年UC Berkeley研究团队在论文中证实，主流奖励模型存在显著的长度偏见，在内容质量相同情况下更长回答系统性获得更高奖励分数65% similarUnverifiedWhen the same model is responsible for both generating content and evaluating quality, a self-referential limitation arises where the evaluation criteria align with the model's own generation preferences, forming a self-endorsement loop.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/11141API
curl https://kongchang.com/api/v1/knowledge/claims/11141MCP
get_claim(id=11141)