Making a hypothetical very powerful AI safe is a pretty difficult problem. Implementing it with a rule of "do whatever the majority of humans vote for" isn't the worst idea possible.

That's a decent arguement against the idea of AI making the world substantially better. But that's not to say things won't change - there are a lot of ways AI could make the world worse.

Why not? Humans could feed starving children. We have the food. It's just expensive to get the food to where it needs to be. AI could find ways to make that more efficient.

Of course that depends on whether the rich people who own AI actually want to use it to feed starving children, but that's an entirely separate question to whether it's possible.

A lot of AI experts predict around a 10% chance of an AI superintellegence causing humanity's extinction within the century.

But of course no one is going to stop working on AI. It's too profitable.

I like my coffee like I like my women. I've never had coffee.

How did the mother buffalo introduce her son who's attracted to men and women?

This is my bison