If AI is programmed to be biased based on any specific ideology/agenda is dangerous. Elon: "The safest thing for AI is to be maximally truth-seeking even if the truth is unpopular... and... it must be extremely curious"