Stanford's Christopher Potts and Peter Hase trace how Chain-of-Thought monitoring became a core AI safety technique

@peterbhase @ChrisGPotts The post: https://ai.stanford.edu/blog/cot-monitoring-history/

this short essay on the intellectual history of CoT monitoring is pretty great!

@StanfordAILab @peterbhase @ChrisGPotts Link?

@StanfordAILab @peterbhase @ChrisGPotts https://ai.stanford.edu/blog/cot-monitoring-history/

@StanfordAILab @peterbhase @ChrisGPotts https://ai.stanford.edu/blog/cot-monitoring-history/

Here is the Blog post

@StanfordAILab @peterbhase @ChrisGPotts always funny seeing safety discourse frame itself as a new discovery

wonder how many past iterations this current wave is unaware of

@StanfordAILab @peterbhase @ChrisGPotts tracing the intellectual history of a research thread is genuinely useful

gives new people a faster way to catch up