Unverified70% confidenceOpinionTime unknown
如果模型能理解行为准则背后的深层原因，即使面对训练中从未遇到的全新场景也能做出合理的行为选择，实现可泛化的对齐
1
Sources
70%
Confidence
Long-term
Relevance
5/31/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAlignment training optimizes the model's statistical behavior in known scenarios rather than establishing formally verifiable security boundaries, and may fail when faced with rare instruction combinations or role-playing frameworks outside the training distribution.78% similarUnverified缩放定律表明模型在特定任务上的表现与训练数据量、模型参数量和计算量之间存在幂律关系，OpenAI、DeepMind等机构的研究支持这一发现73% similarUnverifiedCoT提示词并非教模型新技能，而是激活模型在预训练阶段习得的链式推理能力72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/12269API
curl https://kongchang.com/api/v1/knowledge/claims/12269MCP
get_claim(id=12269)