We’re hoping to alleviate parts of this issue with the use of language models: https://www.lesswrong.com/posts/ebYiodG3MAEqskCDG/a-survey-of-tool-use-and-workflows-in-alignment-research-1
We’re hoping to alleviate parts of this issue with the use of language models: https://www.lesswrong.com/posts/ebYiodG3MAEqskCDG/a-survey-of-tool-use-and-workflows-in-alignment-research-1