Don't develop human-like minds, or prove the stability of a goal system under recursive self-improvement. (If you offer Gandhi a pill that he knows will make hi

Don't develop human-like minds, or prove the stability of a goal system under recursive self-improvement. (If you offer Gandhi a pill that he knows will make him want to kill people, he won't take it, because he doesn't want to want to kill people.) If you throw a bunch of crap together and somehow get a mind from it, it's most likely not going to be anything resembling human (and so you'd be right that not all AIs would reach the same conclusions). I'm not an AI researcher so I'll stop here, but you're assuming that many AGIs are going to even have a concept of "taste" or "dissent" let alone a preference for it.