Claude volunteered to write its findings up as a paper, and recommended that a human number theorist validate its findings
Claude主动建议请人类数学家验证——AI主动寻求外部验证，知道自己可能出错。这种行为比结果本身更值得关注：一个足够智能的系统应该知道何时需要人类背书，而不是盲目自信。这是alignment的具体体现。
1 Matching Annotations
- Last 7 days
-
www.anthropic.com www.anthropic.com
Tags
Annotators
URL
-