I find it so interesting that many people are concerned about AI "waking up" and taking over, which I don't believe can ever happen, but are blissfully unworried about this much more serious and immediate danger: that we ourselves will choose to put machines in control of things they cannot possibly understand. It is the kind of thinking reflected here -- remarkably romantic and naïve for a self-styled "skeptic" -- that exposes us to this danger.

"It might not be such a bad thing if we went extinct" and ceded the world to our machines? For all his classy, educated-sounding British accent, this man is a nutcase.

> For all his classy, educated-sounding British accent, this man is a nutcase.

He just seems to not like people at all, over the years he's gotten more and more aggressive and abrasive when debating anyone about anything. He just comes across as a very unpleasant person.

Assuming AI decided humans were still necessary. Without us it would be a safer place for the planet, but we'd miss out.

After seeing the results of US politics in the last year, the AI would be foolish to keep us around it should be pretty apparent that we cannot be trusted especially en masse.

You think there was a time in US politics that an AI would choose to keep us around?

A silicon based AI might have no use for any carbon based life form. Birds can mess up power lines. Rodents can chew threw cables. I think one of the things AI might do is try to "sterilize" the Earth.

interfering with powerlines and fiber optic cables -- the things an ai would depend on for life -- may be seen as the equivalent of a terrorist attack by the AI.

Edit: further... using silicon or other resources that the AI needs to reproduce or grow would put us in direct conflict with the AI.

perhaps amusingly and somewhat blasphemously, the Bible guides us here: God created mankind in his own image.

Depends on who programmed the AI.

Not even just that - depends on how the training data was gathered

And the choice of objective function.

http://www.smbc-comics.com/comic/2012-04-03

One of the most disturbing things to me about a potential AI future is that anything beyond a very basic education will not be necessary. No point other than personal enjoyment to learn about physics, computer science, engineering, electronics, AI, etc. Not only that, but at some point technology would be so advanced, it may not be possible for a human to understand how things actually work. Everything we know about how our society works will be limited to "lies to children".

Only one way to find out, from the POV of an evolutionary biologist.

better for who.. or what?

Yes, one should always ask: by what standard? surely Dawkins sees that AI would run the world so it gets better for AI itself.

Probably for the sum, or weighted average of happiness of all conscious entities on the planet. This may include ants, dogs, humans, and non-biological consciousness/beings.

If an AI were created to run things, I feel like it would more likely make us stop using hydrocarbon fuels before it would go all terminator on us, as that would take into account our natural right to life right along with all other creatures. That is, if we program it that way. Asimov's The Evitable Conflict from "I, Robot" fits right along with what you're saying about there being some sort of weighted average driven decision process.

Although, I fear we'd probably end up with a "Butlerian Jihad" instead...