Hawking warns of the dangers of AI

[Damelon]

Just an observation, but Professor Hawking has become quite a pessimist in his old age. Didn't he also say that we shouldn't be actively seeking to contact extraterrestrial life: That we find is libel to be far more advanced than us and we risk them coming to squash us.

[Hashi Lebwohl]

We will have much to fear from non-terrestrial species which have attained technology sufficient to reach us physically only if they land here. History teaches us, even when interacting only with ourselves, that when societies A and B interact the one who wins is the one with more advanced tech. If we meet them out there, even if only at/near Mars, then the game changes.

[Vraith]

We will have much to fear from non-terrestrial species which have attained technology sufficient to reach us physically only if they land here. History teaches us, even when interacting only with ourselves, that when societies A and B interact the one who wins is the one with more advanced tech. If we meet them out there, even if only at/near Mars, then the game changes.

I'm not convinced that was entirely true even historically...yea, there is a dominant, normally...and lots of death/conflict...but there is also merging, influencing, fertilization, hybridization.
And I think it matters that we're "expecting" it, in a way.
Also I think that, more and more, the fear of change, transformation, newness [and power/influence of "tradition"] are changing.
If I had to guess, just randomly winging it from general info--100 years ago 90% of any given population [at least] worried...I mean really worried, considered even little things existential threats to all of humanity...about change. Since then I believe a greater and greater share of populations have grown to expect change, to desire it. Heh...in it's way, the longing for "the next big thing" has replaced traditional tradition---and become a kind of "tradition" itself.
[[not totally true, of course...look at how half the country went half bat-shit about ebola...but the arc is bending]]

Anyway, why I came here is there is a site I go to occasionally...cuz the guy is pretty smart, covers lots of topics that I sometimes get curious about...but mostly he just cracks me up. I sent peter to the place once with a link about Grahams number. Today I found he has a part 1 post on AI. It's pretty long, and I know at least some of you have read some of the sources he uses, but y'all might find it fun---and he has little diagrams that often make me chuckle and chortle.
First a quote then the link...the quote is from a section talking about the speed of the human brain:

Ray Kurzweil came up with a shortcut by taking someone’s professional estimate for the cps (calculations per second) of one structure and that structure’s weight compared to that of the whole brain and then multiplying proportionally to get an estimate for the total. Sounds a little iffy, but he did this a bunch of times with various professional estimates of different regions, and the total always arrived in the same ballpark—around 10^16, or 10 quadrillion cps.

Currently, the world’s fastest supercomputer, China’s Tianhe-2, has actually beaten that number, clocking in at about 34 quadrillion cps. But Tianhe-2 is also a dick, taking up 720 square meters of space, using 24 megawatts of power (the brain runs on just 20 watts), and costing $390 million to build. Not especially applicable to wide usage, or even most commercial or industrial usage yet.

waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html

[Vraith]

Yet another person that thinks we're on the road to upload and/or cyborg stuff. A slightly different take/detail in there:

phys.org/news/2015-04-personality-sight-inventor.html

[Zarathustra]

Scientists who aren't worried about robots/AI

[Vraith]

This isn't exactly on topic. But it's reasonably close.
It also connects, to greater or lesser extent, with several other [fairly] recent posts/lines of discussion in a fair number of threads...memes, AI, creativity the future of jobs/work.
It's not a deep thing, just some interest piquing things.

A teaser quote:

it’s important to note how misleading the word technology can be. Most of what our brains impulsively refer to as “technology,” things that have blinking lights or computers inside, showed up only decades ago.

Technology as Kelly defines it, “that which extends the reach of its maker,” is much older; half a billion years at least. Remember, that eagles make nests and beavers build dams. An eagle’s nest is “bird technology.

[peter]

Was there meant to be a link there V? I don't seem to have one.

As an (on topic) aside ( ) I'm finding that my tablet,which is much more up to date than my laptop, has exactly the problem Hawking encountered with his voice box. Sometimes it just doesn't like to be 'overridden' when it decides I've spelt a word wrong or whatever, and I find that when I make the post it has reverted to its version rather than accepting mine!

[Vraith]

Was there meant to be a link there V? I don't seem to have one.

As an (on topic) aside ( ) I'm finding that my tablet,which is much more up to date than my laptop, has exactly the problem Hawking encountered with his voice box. Sometimes it just doesn't like to be 'overridden' when it decides I've spelt a word wrong or whatever, and I find that when I make the post it has reverted to its version rather than accepting mine!

My wife's phone often refuses to be put in its place.

There was supposed to be a link. This one:
singularityhub.com/2015/07/12/can-we-control-our-technological-destiny-or-are-we-just-along-for-the-ride/

[Vraith]

Ok...this is weird link, cuz it's from Reason, which is mostly known for politics-related, especially Libertarian.
It's also half about fiction and half about opinion/speculation and half about science.
And it's a field/topic I SHOULD have thought of, but didn't.
And it both cracks me up...yet deserves some thought. [for implications it contains, and some beyond the scope.]
It could belong at the Onion, an SF convention, or a Future-tech/ethics conference, with some tweaks.

reason.com/blog/2015/08/07/the-menace-of-artificial-stupidity

[Fist and Faith]

I only looked at the first couple paragraphs so far. However, I read the book he's talking about, Avogadro Corp, and the other three books in the series. The AI in AC has an interesting origin, and is not what we usually get in books and movies.

The AI in the second book has, again, an interesting origin, different from the first.

The series resolution is pretty darned cool, IMO. As I mentioned elsewhere, this is your kind of story, Hashi. If you read scifi, that is.

[peter]

It's a chilling thought that our understandings of psychology, neurology and computing could combine to produce algorithms of mass, or even individual persuasive power as to be virtually mind controlling. Or is that already the case........

[Vraith]

It's a chilling thought that our understandings of psychology, neurology and computing could combine to produce algorithms of mass, or even individual persuasive power as to be virtually mind controlling. Or is that already the case........

It both is and isn't the case...so far. As far as we know.
It's chilling...but also potentially awesome. Sure, mind-control [or at least very strong suggestion/manipulation]
But it could ALSO be used for mind-freeing, mind-expanding. To chain, or break chains.

Perhaps the scariest part is getting ahead of ourselves---which we tend to do a LOT. We see/know THAT some method/machine/tool does something, so we start using it, without knowing how/why it does that thing.

Or without realizing [sometimes even realizing/discovering, and not caring] that it is also doing other things.
Much of this stuff is, at root, about influencing behavior even more than thought---if they can get you to act without thinking, that's actually a bonus [at least in the short-term, high impact, instant reward world].
But reward, any reward, is a potential psychological butterfly node: chaotic, complex, complicated, infectious, multi-domain.
[[[it it were not, we would already be living in utopia]]].
The very last thing anyone generating rewards want is an unpredictable explosion/variation. Therefore, they [we?] create a structure of rules, and the force to back it up, to cage or kill off any mutations/expansions we don't like.
I'm not at all sure which is more dangerous in such a case---human intelligence, human stupidity, artificial intelligence, artificial stupidity? It's possible that AI is the LEAST dangerous.

[Hashi Lebwohl]

Human stupidity is significantly more dangerous than artificial intelligence.

[Vraith]

Human stupidity is significantly more dangerous than artificial intelligence.

I don't know...it's a pretty close call. At this point, I'd have to give your statement a bit of advantage because we have plenty of knowledge about human stupidity, but most of what we "know" about AI is primitive/speculative.

Even more topical===there is an "Ask Me Anything" with Hawking coming up, eventually. It couldn't be "live" because of his limitations. The questions are in, though...the ones I saw looked pretty damn good, and the "philosophy is dead" thing is one that will likely get answered.

I'll put up the answers link when they start happening [it might be a couple months, they say].

[peter]

Indeed the [risk is that the] one may lead to the other.

[Vraith]

Here's a guy not afraid of AI necessarily taking over.
More too it than that, about machine learning then, now, and going forward, and I'll probably pick it up.

Excerpts from an article, then link to it:

But the book is not simply an overview, it’s also an argument for the following hypothesis:

All knowledge—past, present, and future—can be derived from data by a single, universal learning algorithm.

Domingos is not talking about creating “revolutionary” and “disruptive” new apps for efficiently ordering pizza or rapidly locating purveyors of craft beer. If his master algorithm is discovered, the hyperbolic vocabulary of tech-industry cheerleading would actually become justified. He predicts that this algorithm would a) Cure cancer b) Eliminate all jobs, freeing everyone to enjoy a life of leisure and making employment just another vestige of humanity’s primitive past, and c) Invent everything that can be invented.

Whether this is attainable is an open question.

devotes a great deal of space and ingenuity to explaining the intricacies of the five major intellectual “tribes” of machine learning: the Symbolists, the Connectionists, the Evolutionaries, the Bayesians, and the Analogizers. Each school of thought has a “master algorithm” of its own, but the ultimate Master Algorithm would combine elements of all five approaches, thus eliminating the drawbacks of each.

www.amazon.com/gp/product/0465065708

[Avatar]

Well, AI is one big step closer:

kevinswatch.ihugny.com/phpBB2/viewtopic.php?p=1010255#1010255

--A

[Cord Hurn]

If Artificial Intelligence can produce the finest craft beer ever made, then I say BRING IT ON!!! I'll worry about AI taking over the world later!

[Fist and Faith]

I also missed your post, Av.

I don't know anything about Go, but here's a thought based only on what I just read.

AlphaGo did what it did because it is not human. That is, with 30,000,000 moves programmed into it, it played thousands and thousands of games with itself. People can't do that. It's not a matter of intuition or imagination that made it able to do that. It's a matter of processing power. So maybe "brute force", processing power, plays a greater role in Go than they're saying it does. Maybe the best players are the ones who can calculate things without realizing that's what they're doing. Kind of like some people can multiply huge numbers in their heads, or tell you what day of the week April 3, 6,543 BC fell on.

What gave AlphaGo this "intuition" other than the processing power that allowed it to experience so much of the game?

[Hashi Lebwohl]

what day of the week April 3, 6,543 BC fell on.

Saturday.

Anyway...computers at this time cannot have "intuition". Technically, "intuition" doesn't exist. What we experience as "intuition" is nothing more than a momentary ability of our brain to process details without conscious effort and without us being required to concentrate on the effort. If we conclude that the probability of an event occurring is greater than 50% then it is more likely to happen than not. We didn't "know" or "feel" that the baccarat dealer was going to flip over a natural 8 for the banker and a 6 for the player; rather, we figured that after the player had won 3 hands in a row combined with the knowledge that the edge on the banker's win percentage is about 2% that betting on banker is more likely to result in success. (the edge also depends upon the size of the shoe but we won't get into that much detail here)

Given that Go is a much more open and complex game than chess it means there should be a higher probability of a human being able to defeat a computer even though the computer is able to process the most likely 1,000 next moves in under a second.

If you want to demonstrate the limitations of a supercomputer you need to ask it simple questions it cannot possibly answer such as "what number am I thinking of?" or "what item do I have in the middle drawer of my desk at work?".