Two models of AI oversight – and how things could go deeply wrong garymarcus.substack.com 1 points by son_of_gloin 3 years ago · 0 comments Reader PiP Save No comments yet.