News for hackers. Stuff that matters.
Re:Why AI is a dangerous dream - interview with Noel Sharkey by gizmo on Friday September 04, 2009 @01:06PM (#805221)
Silly. So very silly.

First of all: he doesn't use his definition of intelligence consistently. According to his usage of the word a being can be truly intelligent, and another one can _appear_ perfectly intelligent but be in reality a dumb machine that merely simulates intelligence. That's his chess example. But he _defines_ intelligence as the science of making machines do things that lead us to believe they are intelligent. A chess computer that outsmarts me most definitely fits that definition of intelligence.

He continually makes the claim that "Technological artifacts do not have a will or a desire". There is no reason to assume that you need some primal force to get a will or desire (perhaps any sufficiently complex system will have a will or desire as a side-effect), and there is no reason to assume a will or desire can't be perfectly emulated.

He claims that there are no signs that computer processing speed will eventually overtake that of the human brain. I say that computer processing speed is undeniably improving rapidly, and that the number of tasks that computers can do is rapidly increasing. Unless there is a hard limit somewhere the assumption should be that we will be eventually overtaken.

Computer AIs can play Chess, Checkers and Super Mario. They can create art, and compose music. They can drive some vehicles, and land planes at night. How about science? Some proofs are made and proved correct almost entirely by computer. Some proofs are so complex they can only be verified by computer. In many research fields a single human is almost guaranteed to contribute nothing. A computer, on the other hand, can probably brute-force his way to many new discoveries.

He again makes the claim that computers are innately unable to feel compassion or empathy. Only to finish with the dumbest remark of the article: "I don't think they will be very good at faking fouls", which is clearly a matter of basic game theory. And I think that robots will completely crush humans at soccer, even if they lack strategy. The moment robots are good enough to take the ball away from a pro human player, they will be able to do so consistently. Even if they run slower and shoot only semi-accurately, they will never make big mistakes. And history shows that in any game where computers can compete the humans have to play (almost) perfectly to even stand a chance (see: Chess / Checkers / Poker). We meat bags with our 100ms+ response times will never be in the same league as robots. Either we will be far superior to the robots, or the robots will run circles around us.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by alexfarran on Friday September 04, 2009 @01:59PM (#805341)
> But he _defines_ intelligence as the science of making machines do things that lead us to believe they are intelligent. A chess computer that outsmarts me most definitely fits that definition of intelligence.

No he doesn't. He defines AI that way, and that's the point. AI is an illusion of intelligence. He sees a danger in mistaking it for the real thing - in effect mistaking machines for people.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by Tichy on Friday September 04, 2009 @02:08PM (#805350)
What's "the real thing"? Of course you could always define AI as "not real intelligence", then by definition AI will never be "really intelligent". It would not be a very interesting claim, though.
Re:Why AI is a dangerous dream - interview with Noel Sharkey by jshen on Friday September 04, 2009 @04:25PM (#805604)
If it's brute force, i.e. chess programs, it's not real intelligence.

If it is a statistical model that produces a result given an input it's seen before then it's not real intelligence.

To me intelligences is the ability to learn by applying patterns/metaphors across domains. Computers aren't anywhere close to this.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacoblyles on Friday September 04, 2009 @07:59PM (#805982)
Applying a statistical model trained on previously observed input is a reasonable description of much of human behavior.
Re:Why AI is a dangerous dream - interview with Noel Sharkey by jshen on Monday September 28, 2009 @03:45PM (#849072)
A chess program trained on previously observed input is not able to then learn how to paint, or drive a car, or play basketball. This is the difference.
Re:Why AI is a dangerous dream - interview with Noel Sharkey by gizmo on Friday September 04, 2009 @02:33PM (#805407)
You're right, he defines AI that way. He seems afraid that robots will take on human responsibilities even though they're unable to feel or understand human emotions. That, I agree, may not work out so well.

I don't see why we should have a different metric of intelligence for computers and humans, though.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Friday September 04, 2009 @01:46PM (#805309)
> A chess computer that outsmarts me most definitely fits that definition of intelligence.

No, it may merely indicate that you are a lousy chess player. What you believe to be intelligent and what is intelligent are two different things.

An intelligent computer would not be programmed with anything specifically related to chess, it would simply learn the game like any human learns board games.

> He claims that there are no signs that computer processing speed will eventually overtake that of the human brain. I say that computer processing speed is undeniably improving rapidly, and that the number of tasks that computers can do is rapidly increasing. Unless there is a hard limit somewhere the assumption should be that we will be eventually overtaken.

That depends on the kind of tasks. Bookkeeping, data storage and retrieval, encoding, decoding, these are all things that a computer is exceptionally good at. But they are all at heart still simply ways to manipulate numbers, basic arithmetic if you want.

Composing a piece of music by computer takes a programmer. No computer ever woke up one fine Sunday morning and wrote a symphony because of the way it felt about the light.

They do not create art, they do not create music, they are excellent tools in the hands of artists and composers.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by gizmo on Friday September 04, 2009 @02:27PM (#805391)
I beg to differ.

According to that logic reading a physics book is "dumb", I should derive all physics knowledge myself. Reading a book is precomputed knowledge, and that's no different from precomputed chess strategies. If I'm allowed to use books as a foundation for my intelligence, then computers are allowed precomputed knowledge also. They're just better at retaining it. Good for them.

The problem with your definition is that you can have a humanoid who speaks a dozen languages fluently, is great at math and physics, can answer any history question, can correctly understand and act in different social situations and despite all those things, is still classified as dumb.

You can define intelligence as the capacity of learning, taking knowledge completely out of consideration. It does not, however, match the definition of intelligence in common language. It also means that when you have a robot that knows everything it is by definition dumb, because there's nothing left to learn.

> But they are all at heart still simply ways to manipulate numbers, basic arithmetic if you want.

And humans are just a bunch of atoms. Atoms form molecules. Molecules form nucleic acids and other building blocks. You can build things of immense complexity with primitive primitives. That something as complex as a human can consist of a few elements is pretty hard to grasp. But if I accept that, then why can't it be simulated with bits?

Art created by computers is primitive, sure. But human art started out pretty primitive too. I see this as a difference in complexity, not as some magic line computers have to cross for their output to become art-worthy.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Friday September 04, 2009 @02:51PM (#805434)
Reading a physics book is dumb, but understanding a physics book is not. You could - given enough time and intelligence - derive all of physics yourself.

Pretty much everybody alive has at some point in his time come up with some rule or law in principle that somebody else had already codified, there is nothing special about that, other than that we seem to have this generalization capability built in at a very deep level.

Your intelligence is what allows you to use the book, I can store the book in a computer but it will not get one tiny little bit smarter because of it. There not being any intelligence present to understand the deeper meaning of the string of symbols that make up the book.

Intelligence in common language is often confused with 'smart', you are either intelligent or you are not. But even the most stupid human is intelligent, and all computers are not.

I can see what you're getting at with the argument that if both humans and computers are built up out of atoms that in theory we could both be equivalent in intelligence.

But there are some very subtle differences between humans and computers. For one we operate in a mostly analog world, and computers are digital. That's a small difference at first sight, because after all if you add enough bits to your floats eventually the steps will be so small that it won't matter any more. Or does it ?

Even a small step is still a step, a thresholding operation.

Simulating a chemical soup, all the effects of things like quantum mechanics, including brownian motion and so on might some day yield an intelligence.

But simply assuming that more computer power will give us something that is intelligent is simply not plausible, more data does not mean you are smarter, just that you have more data at your disposal, a larger computer does not mean thinking in a different way, it merely means following programs and manipulating data at a higher speed.

If intelligence is a continuum between '0' (for a rock) and '1' for a human being with basic intelligence then we have departed a little bit along this axis away from the '0', but we are infinitely closer to the 0 end than we are to the '1' end.

And we seem to disagree on whether or not a computer has ever created art, but that's fine. A computer does not express itself in any way that it has not been programmed to do, and in those cases that it does we reach for the debugger.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by Tichy on Friday September 04, 2009 @02:11PM (#805359)
There are programs that can listen to a composers music and then compose music in that composers style.
Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Friday September 04, 2009 @02:19PM (#805378)
So, effectively it still needed the composer. Imitation is not exactly a great example of intelligence I think.
Re:Why AI is a dangerous dream - interview with Noel Sharkey by gizmo on Friday September 04, 2009 @02:39PM (#805419)
Is it not? When monkeys emulate their trainers that is seen as an intelligent act. Babies emulate everything they can as they're learning. Imitation is a big part of intelligence as we know it.
Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Friday September 04, 2009 @02:54PM (#805437)
I think you miss the point how much programming went in to writing the software that does the imitation. I'm imagining a fast Fourier transform to analyze the incoming audio samples, software that knows about key signatures, tempo and so on. Then given an algorithmic transformation function with a random generator you can come up with new data. But it is really just a transformation of the old data.

Given white noise as input the program would not produce a symphony that would strike us as a masterpiece or move us to tears. That takes a composer.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by petite on Friday September 04, 2009 @03:01PM (#805456)
Composers aren't born composers either. Through years of practice an individual learns how to manipulate the instrument and over this time period, listens to/plays a wide variety of well known musical compositions. Only then is the individual able to use his knowledge to create something new. It would be similar to loading the program with most compositions from a particular time period, and then having something produced which both sounds different enough to be distinguishable, but similar enough to be pleasant.
Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Friday September 04, 2009 @03:08PM (#805468)
Yes, that is true, they are not born composers. But their art is not to take a piece of music that already exists and then transform it according to a set of rules, that is where the 'art', 'inspiration' and 'transpiration' comes in, and it takes a being capable of emotion to be able to do that.

You are seriously underestimating what composing is, it is not merely 'coming up with sounds that are distinguishable and pleasant'. That's the mechanical end of the business, any piece of software that you teach about basic key signatures and tempo rules & phrasing could in theory do that.

Composing is much like story telling, to paint a picture in sound that evokes an emotion with a purpose in the listener.

On the day a computer comes up with something that moves me as strongly as Bach or Vivaldi I'll eat my hat, but I won't be holding my breath for that.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by petite on Friday September 04, 2009 @03:32PM (#805516)
To me, that sounds like saying "and then some magic happens" which ultimately creates the end work. Who is to say that the intermediate steps could or could not be reproduced by a program?

By the way, I'm a painter, so I'm not saying that when I do create a work that it was crafted by a set of logical, binary steps that I always follow for each painting. But I can say that my work can be analyzed to see why I did every single step, and none of those are impossible to express to a computer. Aside from my anecdotal experience, this is something that is also easily observable within the visual arts, and the artists that are recognized for defining the field.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Friday September 04, 2009 @03:38PM (#805532)
I think that in the end it all boils down to motivation and free will, a computer has none, even the ones with random generators.

An intelligent being (artificial or 'natural' makes no difference here) has free will and I believe (but can not prove) that eventually that leads to intelligence.

The prevailing belief here (certainly judging by the moderation ;) ) seems to be that you can algorithmically describe everything. I'm not sure that that is possible, neither am I sure that it is not.

But to date, nobody has designed a machine that on its own decide to go and do anything at all. Even the act of deciding to make a painting vs going out for lunch or talking to someone for an hour or two requires free will and intelligence. No computer in existence today can do that, and if there is not going to be some fundamental change in how computers work I doubt that one will do so in the future.

If you haven't read it yet look at the top of this discussion, rms has posted a link that makes for some very interesting reading on all this.

& thanks for the exchange, it has been educational, especially about the 'method' that could be abstracted from art.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by petite on Friday September 04, 2009 @03:48PM (#805547)
Yes, I understand absolutely where you are coming from here. I simply am not sure whether there exists something inherently different that allows me to make decisions such as eating lunch vs. creating a painting, or whether it's just a matter of complexity within my own memory, (such as my past experience of painting at noon on a sunny sunday has rewarded me with a good painting, which to me is similar to expressing that idea to a program). But again, this is merely philosophical. I just don't want to throw that idea out simply because we haven't done it yet.

And it's been a pleasant exchange on my end as well :)

Re:Why AI is a dangerous dream - interview with Noel Sharkey by Tichy on Friday September 04, 2009 @05:21PM (#805755)
"But to date, nobody has designed a machine that on its own decide to go and do anything at all."

That's simply not true. For example the Google bot decides every seconds if it wants to crawl this website or another one. I don't think you are up to date - might want to read up on "autonomous robotics". Many of the algorithms are so in nature that humans can not predict their results.

I think if you talk to robot builders they will tell you that the robots do stuff they weren't told to do all the time.

As for motivation, I don't think it is as magical as you presume it to be. We are not motivated to do arbitrary things - nature has motivated us to do certain things (think, learn, reproduce). So if we motivate a computer to do certain things (like crawl web site), why should it not count?

Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Friday September 04, 2009 @06:39PM (#805889)
Whether we have the mental capability to simulate any amount of computer time does not mean that the algorithms are not deterministic in nature and that they could be simulated and computed by any turing capable machine, including the human brain. It might take a while, but that's the only difference.

Anyway, I can see I am not going to be able to convince you of this and it is getting late so I cede to you.

But thank you for the interesting exchange.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by gizmo on Friday September 04, 2009 @03:04PM (#805464)
So music isn't music unless it's Mozart?

I don't think it's me who is misunderstanding the discussion here. You're using double standards.

When a primal tribe is beating on drums: music. When a computer builds a primitive melody? That's not music! Those are just bits and bytes!

If I follow your logic to the extreme a computer can never compose music because there will always be a human who built the computer, or built the computer that built the computer that composed the music. I don't buy that.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Friday September 04, 2009 @03:20PM (#805491)
I'm not saying a computer can not compose music, I'm saying that without real intelligence there is no composition, merely the following of a set of rules, a program. Intelligence, artificial or not, is capable of such stuff, a mechanical device interpreting fixed rules is not.

When a primal tribe beats on a drum they've just invented music. When a computer beats on a drum it has been told by some human to beat a drum.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by Tichy on Friday September 04, 2009 @04:49PM (#805667)
"When a computer beats on a drum it has been told by some human to beat a drum."

Well yeah, it seems likely that if a computer ever becomes really intelligent, it has been built by a human. So?

Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Friday September 04, 2009 @04:57PM (#805689)
If it is an intelligent computer it will not need to be told what to do, it will motivate itself, but we may not even recognize its actions as motivation because the concepts are not clarified, we only know what motivates us, we can not begin to imagine how a sentient AI would be motivated, maybe it will tell us, maybe it won't, who know... That motivation does not necessarily have anything in common with a human motivation, there are plenty of alternatives. It might not want to beat a drum, even when told to.

For that to happen though we have a bunch of formidable hurdles to cross, and in spite of all our processing power and data storage capability we have not yet been able to simulate the intelligence of even an insect or a small mammal.

Whether an AI is built by humans or aliens or comes in to being through some freak coincidence is irrelevant.

As long as it takes a human with their intelligence to tell a computer what to do it has not achieved intelligence, merely the capability to follow instructions very precisely.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by kd5bjo on Friday September 04, 2009 @07:31PM (#805953)
Are these human composers not "merely following a set of rules", namely quantum mechanics?
Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Saturday September 05, 2009 @04:09AM (#806338)
You're in line with Roger Penrose here, who suggests that in order to generate consciousness we will have to look at quantum effects rather than at molecular level effects.

This is possible, but not a certainty.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by kd5bjo on Saturday September 05, 2009 @01:23PM (#806787)
From what I understand, molecular level effects are the same as quantum effects looked at from a slightly higher level. My point was that, as far as we can tell, everything people do is the ultimate result of applying a fairly simple set of rules (the physical laws).
Re:Why AI is a dangerous dream - interview with Noel Sharkey by jacquesm on Saturday September 05, 2009 @01:34PM (#806801)
I don't think it is quite that simple. Quantum effects can screw up the apparent 'rules' of the macro world (atoms, molecules and upwards) subtle the closer you get to the lower levels of the macro world.

Physics as you know it ('classical physics') stops making sense at that level, things no longer have an apparent causal relationship.

What Roger Penrose argues in his book 'the emperors new mind' is that 'mind' springs from these effects.

He suspects that if we simply follow the classical physics rules that we will end up on a dead end and that it will take a full blown quantum simulation of a whole brain in order to simulate one with the emergent behaviour that is being sought.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by kd5bjo on Sunday September 06, 2009 @11:30AM (#807873)
Certainly. Setting aside whether or not quantum (instead of classical) physics is necessary, it certainly seems to be sufficient to explain everything that a person does. Given that, how can a person not be considered to "merely follow a set of rules"?
Re:Why AI is a dangerous dream - interview with Noel Sharkey by Tichy on Friday September 04, 2009 @04:45PM (#805657)
Human composers also learn by listening to works of other composers before them.
Re:Why AI is a dangerous dream - interview with Noel Sharkey by anigbrowl on Friday September 04, 2009 @06:21PM (#805865)
Oh God.

We can automate a process, and with technology we can automate increasingly complex processes. But intelligence is a fundamentally different problem from the things you describe. I am a believer in the possibility of strong AI. I can envision and would like to live a world of sentient machines. But your way of thinking, as expressed here, will not get us there.

Let's take the example of robot soccer players. I agree with you, it's basically just an engineering problem. We have robots that walk around, we have computers that can solve ballistic problems in real time, we have networking and so on. I can easily, easily imagine a winning soccer team made up entirely of robots. If I was rich I would tempt you with money to hasten the creation of such a thing.

What I can not imagine - or rather, what I am unable to conceive a route towards - is a robotic soccer player that announces one day it is sick of putting balls into goals and is retiring to take up painting instead.

Or take chess. I can't beat my computer at it any more, although that's not saying much as I was never all that good. Yes, I can envision a chess program that says 'hmm, you keep losing, how about I start without a pawn until you improve?'. Indeed, we're not so far away from being able to build a good chess tutor, not least because chess play has a fully defined grammar and so winning positions in chess games could be regarded as true statements in a formal system. On the other hand, further development in this area is unlikely to produce software with a secret addiction to backgammon.

We have software that can create pictures or music in a certain style. But although they can produce interesting and unexpected results they're still horrendously limited in other ways. If you request a new composition and don't like the results, will your computer denounce you as an ungrateful philistine, while mocking your bourgeois notions of formalism?

Computers don't want to do things. We want them to do things, and currently all we know how to do is build simulacra of ourselves. Doubtless, as time goes on, we will be able to build more general-purpose simulacra - our robotic soccer players will also compete at rugby and tennis, our chess programs will handle a wider domain of dynamic analytical problems, our artistic software will employ inductive processes that begin with general concepts of proportion, contrast and so forth.

But we are not going to get to real artificial intelligence until we start building systems that are a) self-modifying and b) allowed to engage in perverse behavior whose purpose is not obvious to us.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by naveensundar on Friday September 04, 2009 @03:52PM (#805555)
> And history shows that in any game where computers can compete the humans have to play (almost) perfectly to even stand a chance (see: Chess / Checkers / Poker).

What about GO? http://en.wikipedia.org/wiki/Computer_Go

Re:Why AI is a dangerous dream - interview with Noel Sharkey by rwolf on Friday September 04, 2009 @04:28PM (#805610)
The important part of the quote there is "where computers can compete". Go-bots have a long way to go before they'll be considered competitive.
Re:Why AI is a dangerous dream - interview with Noel Sharkey by naveensundar on Friday September 04, 2009 @04:44PM (#805650)
What do you mean "where computers can compete"? That is just begging the question. Is it ok to assume Turing based computation will go where you say they will go?
Re:Why AI is a dangerous dream - interview with Noel Sharkey by xiaoma on Saturday September 05, 2009 @07:32AM (#806420)
It means where computers are competitive with even a journeyman club player.

4 years ago, I would have said,

Computers "can compete" at go about like humans "can compete" with Siberian Huskies in multi-day footraces over snow-covered tundra. I'm a terrible go player, and yet I can still give the strongest go program in the world full 9-stone handicap and crush it.

Interestingly, more computing power hadn't really helped. The best programs of 2005 sucked just about as much as the best programs of 1995. Recently, though, it's changed. MoGo beat a 9-dan professional with just a 7-stone handicap this year at the 2009 Taiwan Open. And it's won multiple games against pros on smaller boards straight-up.

http://epochtimes.com/b5/9/2/11/n2425091.htm

Re:Why AI is a dangerous dream - interview with Noel Sharkey by naveensundar on Saturday September 05, 2009 @10:45AM (#806570)
Still begging the question using fuzzy terms. "If computers are sort of good they are very good!" That statement doesn't mean anything concrete. Don't the recent results in GO invalidate : if(computers can compete) {humans have to be perfect} ? When does the conditional get satisfied? The moment there is a basic implementation which follows all the rules of the game or according to some fuzzy notion of "compete" which gets changed conveniently :)
Re:Why AI is a dangerous dream - interview with Noel Sharkey by xiaoma on Saturday September 05, 2009 @06:09PM (#807115)
In the simplest terms, it means that AI performance is less variable. A computer that can give you a good match on a good day will beat you on the days you're tired, distracted, or ill. And the computer won't have bad days.

Conversely, if you're a bit better than the computer and manage to find a few of its weaknesses, you can exploit them again and again and it will never learn how to stop you (in any AI I've played against).

Re:Why AI is a dangerous dream - interview with Noel Sharkey by rwolf on Saturday September 05, 2009 @10:54AM (#806576)
The poster you replied to did not say Go-bots were competitive yet. He said they had shown a lack of progress in the face of increasing resources until recently.

The only thing I'm saying here is that your counterexample does not make sense because Go-bots are famously bad at Go.

Re:Why AI is a dangerous dream - interview with Noel Sharkey by naveensundar on Saturday September 05, 2009 @05:47PM (#807081)
>The only thing I'm saying here is that your counterexample does not make sense because Go-bots are famously bad at Go.

I don't think so. Please see the link below: http://en.wikipedia.org/wiki/Computer_Go#Recent_results

Of course you can still refuse the recent results, and say they are not competitive :)

Re:Why AI is a dangerous dream - interview with Noel Sharkey by rwolf on Friday September 04, 2009 @10:08PM (#806067)
he said if (competitive) { requires perfection from human opponent }

you brought up a game where the condition does not apply. "competitive" in an understood word: can a computer tie consistently against what we agree is a moderately skilled player?