next
ultimately the question is not if we should trust AIs, but rather if AIs should trust us