its simply really. every time we run through the loop where the AI makes itself smarter we also ask it how to align AI. presumably as its power diverges the AI will converge to the correct answer

Posters: eigenrobot. eigenrobot

2 likes recorded by the source corpus.