Elon Musk claims humanity will only survive so-called AI Revolution with 5-10% chance

UsedToBeKnowItAll wrote:

Unfortunately this is true, and Musk is correct. Our current engineers have so much hubris, and very, very few people comprehend the power of modern computers.

The only part that seems unlikely is that an AI would target our nuclear arsenals. It will come up with a more creative solution to either enslave or destroy us. One that we haven't even imagined. And it will execute it in seconds or less. We wouldn't stand a chance.

I think Google is being pretty reckless in this area of research.

There is a lot of sci-fi about machines taking over the world. The fiction in those movies is that we might be able to fight back.

To the first bold part, agree completely. The vast majority of humans are completely ignorant of the power of the technologies that are now the glue of the every day lives of most people, and we're only at the foot of the mountain.

To the second, might it be that a sufficiently advanced AI might not even consider us at all? I read a very interesting article once conjecturing that in the first nanoseconds after technology achieves something like 'self-awareness' (whatever that would look like; certainly having a 'soul' as we think of it would not be a requirement and the fact is we don't really understand at all the nature of our OWN sentience...) it will be so far past us as to render us like cats or dogs in comparison...

The fact is, it is hubris on a colossal scale for anyone to consider that we are somehow this apex of evolution; mostly I find this in religious types who believe 'gawd' created us in his/her/its image and that we are apparently the center of the 96.x billion light years of the observable universe, when in fact we are simply somewhat more intelligent apes just advanced enough to completely despoil our nest, build the weapons of our own destruction, commit genocide on a scale of millions of our fellow humans, and on and on. We have been around for only a very short time relative to the most successful species on the planet, and we are continuing to evolve (albeit very slowly) as I sit here and write this...

But all of a sudden along comes this thing of our own creation, evolving at hyperspeed in response to environmental stimuli (in this case our consumer demand, demand for advanced weaponry, scientific curiosity, etc) and it is going to far outstrip us, and that horizon isn't even that far away really. We may think of 100 years as a long time but it's not even the blink of an eye in universal time, but 100 years ago flight was still in it's infancy and computers didn't exist at all, and look where we are now. With information sharing on a massive scale where will this all be in 10 years, 100, 1000??

It isn't us that will solve the problems of interstellar space travel, it will be machines, and us? Home sapiens as we know ourselves now may or may not be around in 100,000 years but if we still are we would look upon the humans of today as we look back on Neandarthal or Cro-Magnon man.

The limitation of most people is they just can't extrapolate their own little day to day existence into something deeper, whether looking back at geologic 'deep time' or forward into the future; 500 years ago one might have lived in the same village all their lives and day to day life would not have changed appreciably. Now, life changes dramatically nearly on a yearly basis but there still seems to be that 'static' view of life held by people that somehow we are always going to be (or are SUPPOSED to be) the center of it all when in fact nothing could be further from the truth. Now the game changer of our own creation is here...

Musk can't even accurately predict the output of one of his factories.

Only 5-10% of the population will survive when the roads are filled by 'self-driving' lithium bombs.

Killer Robots are coming wrote:

Tesla's Elon Musk warns we only have 'a 5 to 10% chance' of preventing killers robots from destroying humanity

Elon Musk was speaking to employees at his firm, Neuralink, this month

He said that efforts to make AI safe only have 'a 5-10% chance of success'

The warning comes shortly after Musk said that regulation of AI was drastically needed because it's a 'fundamental risk to the existence of human civilisation'

http://www.dailymail.co.uk/sciencetech/article-5110787/Elon-Musk-says-10-chance-making-AI-safe.html

5-10% against the robots is pretty good. We have 0% against ourselves and no doubt will cause our own extinction.

Elon Musk wants to put young Thai boys into his metal phallus.

That article linked about Musk going to Thailand with a mini-sub is sensationalism without any substance. His actual posts have no snark or criticism in them, yet the article uses them as a basis for calling his behavior petulant, etc. Give it a rest.

I appreciate that he is always going for the tech fix and largely creating it himself, though I wouldn't want my kids to be the guinea pigs on the test trips.

Bad Wigins wrote:

CuriousPerson wrote:

From the former MIT AI director:

(all embedded-link quoters must be band)

Silicon valley billionaires don't tend to be all that smart (outside of the Google founders); they're basically successful salesmen. Don't take things they say seriously.

reason #3 is the most important, but he doesn't mention that it's not only impossible to build a "competent" AI as opposed to a specialized one, it's also impossible to build a truly effective specialized one either. Even if it's rooted to its own sandbox and doing one thing, every decision you give it to make will create dozens of new things that it can do wrong in ways you never predicted. Specialization eventually requires complexity, unless you're building an automated oxcart - or maybe one of Musk's new simulated factory relays.

I like your use of the word, "impossible."

tee hee

Ever heard of PayPal or The Boring Company?

AI has only been in robots for four years.

You should worry more about Talos, the bronze robot built by the Cretins four thousand years ago; it circled Crete three times a day, guarding its borders. And who was it that notified the Athenians that the Persians had landed at Marathon? Today I found out that the U.S. Navy bought some Talos, and they can fly!

True AI doesn't exist in any form yet, so it's very hard to project the course of its development. It may require something like mapping an entire neurological brain with software, which is something that we're a very, very long ways away from being able to do.

If by AI we're simply referring to software that can solve problems, like win at various games, then "AI" is just an arbitrary term to describe software that's gotten pretty good at certain things. But the kind of "AI" that's being developed now is fundamentally different in kind from the AI that would try to destroy the human race. We could get "thinking" software that's 10,000 times more powerful than what we have now, and it would still be a mindless tool.

As another poster pointed out, the kind of AI that we have now might end up being an incredibly powerful mindless tool that people could use in warfare or something. I'm not especially concerned with that, however, because humans have had the technology to wipe out the human race for a long time. And while it's true that the barriers to entry with AI would likely be lower than with hydrogen warheads, there's good reason to think that other AIs could be good countermeasures. In any event, humanity has become progressively more peaceful throughout its history, and I would expect that trend to continue. Especially as technology reaches the levels that we're talking about, human prosperity it likely to continue climbing, which makes conflict less likely.