zlacker

That is a secondary and huge problem, but the larger initial problem is making sure the AI aligns with values that nearly all humans have (e.g. don't kill all humans)