Why not? I'd say that a world that is managed by AGI with limited input from human beings is a good goal to have. If AGI could be done without the nasty parts of human psychology and they're inherently superior to genetically intact human beings why shouldn't be embrace it?

I understand that it's a big assumption to make -- that a benevolent AI could be constructed. But under that assumption, why not have a benevolent dictator in the form of an AI?

> Why not? I'd say that a world that is managed by AGI with limited input from human beings is a good goal to have.

We already live in that world, with large institutional bureaucracies playing the role of paperclip-maximizing AGIs.

It's pretty wretched when you are in their path.

Maybe. Yeah, human politics and justice systems leave something to be desired. But my worry was little bit beyond that. That the AIs would take all the meaningful work, discoveries and creativity away from us, leaving us just to amuse ourselves. Some people might be okay with that, but I don't think becoming pets is the best goal for the human race.

If the benevolent AI ruler(s) restrained themselves to allow for humans to flourish, then okay. Assuming it could be constructed benevolently.

There is another threat when things go wrong (and they eventually always do) - no matter how horrible some dictator is, eventually he/she will die, and at some point things get reshuffled by war/revolution/some other more peaceful means.

With AI, it would try its best to preserve/enhance/spread itself forever. And its best might be much better than our best...

Well, _we_ don't really play nice amongst ourselves, so my retort to you would be:

How much worse could it be?

If Skynet determines we're the problem (wars, famine, global warming, inequality, non-cooperation etc), I'm losing counter-arguments by the day.