待验证50% 置信事实时间未知
RLHF is intended to align the model with human preferences, but human preferences tend to reward 'confident and detailed' responses, creating a paradox that reinforces hallucination
1
来源数
50%
置信度
中期 (~90 天)
时效性
2026/7/2
首次发现
有效期至：2026/9/30
来源
相关事实
引用此条事实
Stable URI
https://kongchang.com/claim/54997API
curl https://kongchang.com/api/v1/knowledge/claims/54997MCP
get_claim(id=54997)