The Dangers of AI?

jerusalem90

Mon, Apr 17 2023, 11:28 pm

I think about it this way: we are much much smarter than ants. Do we kill ants when they are inconvienient to us? Yes, definitely. Do we have any desire to kill ALL the ants in the world, or to intervene in their political affairs? None at all.

So yes, I believe there is a danger, but I think it's weird to assume that just because you are much much smarter than someone that the next logical step is "let's completely annihilate them!"

If you think about the Europeans coming to America, totally totally different. For one thing there was competition over physical resources (land, etc), which I don't see being relevant to AI. For another, Natives were less technilogically advanced than Europeans so while they certainly had a disadvantage they were still a huge threat, could and did kill Europeans. If AI are smart enough, that won't be the case, and they might even intervene only to keep humans from nuking each other, etc, assuming a nuke would be bad for them too. Similarly, if an earthquake, tsunami, etc would be bad for them (I'm assuming they have hardware somewhere which could be damaged/destroyed), they may alert us to those.

On the other hand, they may be interested only in math or music and spend all their time and resources on that and not care about anything else.

Or, they may become "depressed" and commit "suicide".

There are a lot of options, but I think they assumption "they will want to completely annihilate humanity!" is one of the less logical ones.