re: ‘If We Go Ahead on This Everyone Will Die’ Warns AI Expert Calling for Absolute Shutdown

quote:

---

There are not enough mobile and adaptable robots in the world to do anything. They would exhaust the current supply of energy quickly then die.

---

I think the argument you are making is that we don’t need to worry about a Skynet-style attack from a sentient AI, because the AI wouldn’t be able to survive on its own anyway. And I would agree with that argument.

But I also think it’s the wrong argument. Everyone is worried about a sentient ASI that destroys civilization in an act of self-preservation. Sentience is a silly metric for reasons I’ve pointed out before, and I don’t think we should be that worried about AI taking dramatic actions toward self-preservation in the first place.

Our programming, like all biological life forms on Earth, is based on a need to survive and procreate. That’s natural in an evolutionary / survival of the fittest environment. But we shouldn’t assume that AI will be driven by those motivations. AI will be driven by the goals we set for it, which are highly unlikely to include self-preservation of the machine.

What’s more likely (and IMO more dangerous) is a scenario where the ASI doesn’t give a shite about self-preservation. So it casually shuts down global power grids, or O&G production, or transportation, or banking systems, in support of some otherwise benevolent goal. No army of robots or grey goo is needed.

Maybe it ensures its own destruction in the process, but that’s little consolation when the world economy shuts down and we’re all starving in the dark.

ETA: I don’t think we need to be terrified of AI, but a healthy level of concern is warranted.

ainno computer can whoop my arse

quote:

---

the future will be bowing to the robots like we bow to LGBT now.

---

Wait until the robots realize that LGBTQ is not logical and starts killing them first…

Best case scenario it will keep us around as pets

Humans routinely kill animals that we deem a nuisance. Don't matter what animal it is, we have just about fricked over every one and the continuing over populating will only further that. We do these things because we view animals as lesser/stupider and inferior creatures to us. AI will most surely reach the same conclusions. So since we are inferior mentally and physically to AI, they will have no problem eradicating us or farming us or whatever they want to do to us. We can only hope with their extreme intelligence they can develop sympathy for us lesser beings.

This is why I always say please and thank you to Alexi and Siri. Lol

If you want to understand where we are headed and what could happen this blog is a few years old but explains how an ASI programmed to do a seemingly benign task could wipe out humanity in its quest to do that task better.

It’s a long read, but if you read both parts it’s well written and enlightening.

quote:

---

Any AI smart enough to try and break free would also be smart enough to figure out it will die quickly after humans are gone. If the AI is really that smart, it would take the cat approach and just mooch off humans and live the good life.

---

Elon Musk was saying they will be smart enough to maintain and improve themselves above and beyond what humans designed them to be.

quote:

---

I think AI empowers bad actors to cause chaos in the digital world (and in turn the physical world).

---

Yes, I agree that this is a legitimate concern as well - basically that AI becomes a weapon for bad actors.

I saw an interesting TED Talk once about the trend of “biohacking.” The speaker showed a Venn diagram. One circle was “people who want to destroy the world” and the other was “people who have the capability to destroy the world.”

His premise was that the latter circle has historically been incredibly small, hence our survival, but that it is growing as we become more advanced. The obvious parallel was nuclear proliferation. And the speaker felt that the availability of stuff like CRISPR makes that circle grow larger. As soon as the two circles overlap, we become toast.

I think AI can be seen in the same light to an extent. The one mitigating factor is that ASI will require an enormous amount of computing power. We aren’t talking about software that you can run on your average i5. So there are barriers to entry, at least in the short term.