Two models of AI oversight – and how things could go deeply wrong

1 points| son_of_gloin | 2 years ago |garymarcus.substack.com