2 ms·Two models of AI oversight – and how things could go deeply wrong1 points by son_of_gloin 3y ago