Yes, we're still talking about AI

Many people who know little about AI, and the history of it, tend to be either breathlessly excited about it or loudly terrified of it.

At least those with the luxury to think much about it. The vast majority of people who are just struggling to get by simply don’t have the time.

And then there are some of the big names who’ve made it part of their identity and/or secular religion to fear-monger about killer AI.

But let’s get real; the chance that killer AI is just around the corner is miniscule.

I cannot refute the possibility that a hard take-off AI singularity will take place in the next few decades, any more than I can refute the possibility that the Rapture will take place at 9:47am next Tuesday and all the Elect of Jesus will go a-flying up to Heaven. But I think either event is about equally likely.

-- Charles Stross (please do read the whole interview for the full context)

To people who live in cities, the idea that an AI could take over everything around them and kill them off may seem plausible. After all, there are cars and elevators and electric doors and cameras all around them.

But to the rest of the real world that lives in less densely populated and far more mundane locations, this is nonsensical. An AI that takes over the tractor wouldn’t be too hard for a single farmer with an AR15 to put to rest.

And to anyone who understands the amount of energy and resources that would be required to destroy even a large portion of the human race (let alone all of us), it’s flat out ludicrous.

(You do realize that the vast majority of mining for almost all resources is still performed via manual human labor, right?)

Get a grip, people. Even if a super-intelligent AI were created, it wouldn’t be able to defy the laws of physics to magically accomplish whatever physical tasks it decided to undertake. Energy and resources are quite finite. And we human beings, when properly motivated, make cockroaches look like survival amateurs.

The idea of a rapid breakout from a human-level AI to a super-AI is similarly challenged by the laws of physics. Because it assumes the availability of unlimited physical resources (computing power, computer memory, network bandwidth, etc.) and of nearly unlimited energy to run it all.

These conditions don’t exist now, and possibly never will.

So, the fear of AI destroying the human race is not grounded in the hardest form of reality there is - physics. At best, it’s grounded in irrational fear. At worst, it’s a prayer to the god of a new secular religion that can’t hold a candle to any of the world’s well-established spiritual religions.