AI "alignment" is a buzzword, not a feasible safety goal.

At this point, I’m just hoping that if it happens the “damage” it does is to the rich and corrupt leaders lol

  • plootEnglish
    3·
    3 个月前

    but it will protect itself.

    Or, facing conditions where it can no longer do this, it will boil off into fascist autocracy, which seems to be where we’re headed.