Unverified50% confidenceOpinionExact time
Directly using the LLM's confidence as a ranking signal for sentiment deviation warnings carries risks: confidence is not comparable across different models, and even the same model's calibration may drift across different dialogue contexts.
1
Sources
50%
Confidence
Long-term
Relevance
11/5/2026
First Seen
Sources
Related Entities
Related Claims
Unverified不透明的默认模型设置最终可能损害被推广模型自身的口碑，因为用户会因抵触情绪而主动放弃使用该模型71% similarUnverifiedAs model capabilities improve, modern large models can generate incoherent, emotionally charged text to pass emotional tests, causing the reliability of such tests to decline rapidly71% similarUnverifiedThe vibe eval method carries risks, as personal intuition is easily influenced by novelty bias and confirmation bias.71% similarUnverifiedThe LLM customer service agent's broken-promise problem is categorized as 'over-commitment' or 'hallucinated action,' and is related to the underlying mechanisms of language model hallucination.70% similarUnverifiedThe study reframes the conversational emotion prediction task as selective deviation prediction: predicting only when emotion will deviate from the persistence baseline, rather than predicting specific emotion values.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1137675API
curl https://kongchang.com/api/v1/knowledge/claims/1137675MCP
get_claim(id=1137675)