Unverified50% confidenceFactExact time
Standard reinforcement learning agents tend to rely on spurious correlations, while large language models are prone to hallucinating game rules that do not exist.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAutoregressive language models have never seen 'I don't know' as a high-frequency reasonable output during training, so they are more likely to generate fabricated content at the boundaries of their knowledge than to express uncertainty73% similarUnverifiedThe safety behavior of large language models is a statistically learned tendency rather than hard-coded rules, and cannot provide mathematical safety guarantees.71% similarUnverifiedThe Rule mechanism has an inherent limitation: rules are essentially still natural language, and AI may forget them in long contexts due to attention dispersion71% similarUnverified多智能体强化学习领域的非平稳性问题在多人世界模型中被放大，因为其他玩家的行为分布会随策略动态改变71% similarUnverified反事实问题能有效区分真正理解与模式匹配，因为这类问题几乎不可能直接出现在训练语料中70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/893469API
curl https://kongchang.com/api/v1/knowledge/claims/893469MCP
get_claim(id=893469)