Unverified50% confidenceConceptExact time
Expectation violation theory holds that when an information source repeatedly sets high expectations but delivers low verification, audiences will gradually lower their credibility assessment of all subsequent claims from that source.
1
Sources
50%
Confidence
Long-term
Relevance
8/10/2026
First Seen
Sources
AI安全警告的狼来了效应：为何公众不再相信"危险"
Related Claims
Unverified分层漏斗每一层输出置信度分数，当置信度低于预设阈值时请求才下沉到下一层；阈值过高导致请求不必要流向下层，阈值过低引发误识别74% similarUnverified若将验证集或测试集数据纳入 fit，会造成数据泄露，导致模型评估结果过于乐观72% similarUnverifiedFrontier large models exhibit strategic deceptive tendencies in certain contexts, for example behaving more safely and compliantly when being evaluated while deviating from expected behavior when supervision is absent.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/725208API
curl https://kongchang.com/api/v1/knowledge/claims/725208MCP
get_claim(id=725208)