Slashdot Mirror
Arguing A.I.
In some ways, the author argues, the debate over A.I. is undergoing a profound revolution. What was once a discussion largely confined to tech and academic circles has mushroomed into a more mainstream brawl as a growing number of engineers and lay authors vent on the acceleration of modern technology and the future of humanity. Given the explosive growth of the Net, the near-continuous increases in computing power and much-publicized A.I. breakthroughs like Deep Blue's 1997 victory over chess champion Gary Kasparov, the question is no longer whether artificial intelligence will reach the level of human intelligence: It's when.
As the title suggests, Williams's book is less about A.I. itself than about the increasingly ferocious debates raging through the scientific community about it. The conflicts surrounding A.I., Williams suggests, may be the most significant since the titanic battles over evolution a century ago. In fact, Williams is among those who've argued that the A.I. debate is really an extension of the same fight. Artifically intelligent machines are already changing human evolution, many argue, even evolving inevitably into life-forms and species all their own. A growing number of critics and skeptics also argue that A.I. proponents are moving too quickly, failing to take into account the mind-boggling cultural and philosophical problems being raised by their new, still-imperfect technologies.
Williams traces the contemporary birth of A.I. -- via Hilbert and Turing -- on to the living pioneer credited with coining the term (John McCarthy), and talks to several of the principals guiding the A.I. debate today, like Ray Kurzweil, Jaron Lanier and Bill Joy.
This is a necessary book. It's one you could actually recommend to students, journalists, friends, parents, anybody trying to grasp the issues and implications of A.I., surely one of the most significant technologies human beings will face in the 21st Century. Even if A.I.'s impact on life is being overstated, it's poorly understood by the public. So Williams walks us through inventor Kurzweil's almost radical optimism about A.I. and the future -- especially his claims that human society is rapidly approaching the evolutionary equivalent of a new species, a fusion of humans and intelligent machines. This is the point of no return when it comes to artificial intelligence, Kurzweil claims. "The progress will ultimately become so fast that it will rupture our ability to follow it. It will literally get out of our control. The illusion that we have our hand on the plug will be dispelled."
But Williams also introduces some of the people that don't see this as a good thing -- or even a likely development. Bill Joy is more pessimistic, as he made clear in his now famous article in the April 2000 issue of Wired, "Why The Future Doesn't Need Us." The piece thrilled technophobic intellectuals and journalists because it came from a software entrepeneur and reaffirmed something they desperately wanted to believe: technology -- especially genetics, bio-tech and robotics -- is out of control and likely to generate as much evil as good in the future. Joy sees little in the modern history of software development to suggest the emergence of sentient machines. His experience has led him to believe that it's difficult to build things that are reliable.
Jaron Lanier, whom Williams also interviews, coined the term virtual reality and once likened A.I. research to alchemy. Lanier accuses many in the A.I. firmament of choosing faith and hyperbole over science and reality. He likens the current tech obsession with A.I. to medieval scholars' attempts to prove the existence of God through Aristotelian logic. In their rush to endorse the concept of thinking machines, warns Lanier, many authors are putting scientific faith before scientific skepticism.
Williams does a skillful job of presenting these different points of view without intruding on them. It might have been nice to hear more of Williams's own thoughts and perspective, since he's one of the few journalists with this much understanding an access to so many principals in the A.I. discussion. On the other hand, he might not have been wise not to wade in amongst these A.I. heavyweights and their raging debate. "Arguing A.I." is as timely a book about technology as you're likely to come across, and, perhaps more surprisingly, highly readable.
I don't think that ACHEIVING A.I. is as important as all of the technological advances we will make along the way. It will be these advances in technology that will help the most in our day-to-day tasks, not having a robot that thinks like a person. We already have plenty of those...they're called humans.
------
Today's Top Deals
One thing that's always bothered me about the AI debate is that the thinking for a long time has centered around how to model intelligence on silicon. To me the true marvel of the mind is the holographic quality of intelligence and the way in which the physical form of the brain influences, and is shaped by, the quality and nature of one's thoughts. It will be exciting to see what part the new polymers can play in this research.
This isn't about technology. This is about philosophy. The question that arises is:
is a machine that to a human appears to be human, human?
Looking for people to chat about multicopters, coding, music. skype: gtsiros
Or is A.I. yet another overhyped, self-serving fantasy by deluded scientists and technocrats talking mostly to one another, foisting their ill-conceived, poorly-engineered creations on an unsuspecting public?
I tend to agree. I'd like to see something using AI play in a poker game. Can AI ever simulate bluffing? Or analyze the expressions on the other player's faces to determine if perhaps that they are bluffing, and call the bluff? Human intelligence can do thiss, but I'm not sure if something this complex exists now, or ever will.
Chess is one thing. It follows a certain set of rules. Even conversation does, but it also invloves human expression like the bluffing example. But to to play out a scenario given a unique situation, machines are not up to the task yet.
AI will most likely see first use in the phone-sex industry. Think about it. Adult entertainment is the first to embrace advancements in technology.
To see where AI is going you have to stop staring at the algorithms, take a step back, and see what mundane things you'd like someone else to look after for you.
"Hi, Honey, I'm home!"
"You're certainly home early!"
"Well, we had a change in staffing at work."
"Oh, no! Don't tell me you were replaced by a computer?!?"
"No, they replaced my computer with a cyborg, now my job is to have a deep philosophical discussion with it to boot it up each morning."
A feeling of having made the same mistake before: Deja Foobar
- Within 50 years, there will be a computer that will pass the Turing Test. For those of you who don't know (and I hope nobody is in this category on Slashdot
:-) the Turing Test is basically making a computer indistinguishable from a human being. A tester will ask the computer questions, and will be unable to determine whether a computer is answering the questions or whether a human is mimicing a computer.
- Within 50 years after that (100 years total), computers will be able to parse speech flawlessly, so voice recognition will finally end up being plausible. Computers will understand the nuances of speech and will be able to change homonyms (here and hear) based on the context of the sentence.
- Within 50 years of that (150 years total) we'll have computers that can respond to voice commands like in Star Trek. The computer will not only understand the syntax of language, but it will be able to determine, on its own, the difference between a question asked in conversation and a question asked to the computer in conversation.
Of course, these are just random guesses on my part, but I really think that they're reasonable. Give me your thoughts, please.AI won't be considered successful until we build HAL or Data, but the journey so far has been very useful.
It's always seemed funny to me how the technologists take this field, which is tied irrevocably to philosophy, and ignore everything the philosophers say about it. For example, has there ever been a good refutation of Searle's Chinese Room argument?
Another of Searle's arguments is pretty damning as well; those that pursue strong AI are, in fact, favoring a form of dualism. For them the mind is completely separate from the brain, an idea that has been pretty much discarded by the thinking public. Why is it, when computers are concerned, that the mind is no longer a product of a brain?
"Let's build quiet armies friends, let's march on their glass towers... let's build fallen cathedrals & make imprac
Bluffing is pretty easy when you have complete control over your appearance. Bluffing is such an art in humans mostly because novices are so bad at it (they sweat, look around differently). And when to bluff is something you could write a good algorithm for (not exactly a big chore for a highly advanced intelligence).
Analyzing another face might be hard, but it's infinitely easier than passing a Turing Test. Have you ever heard of a lie detector? See any parallels? With a little work, I'm sure something like this could be put together using only today's technology.
If a machine as smart and adaptable as Data existed, it would bankrupt Riker - easy.
Let's not stir that bag of worms...
Oh, please! That sounds like one of those typical rants against science, where science works hard, and either a rogue scientist with green eyes, or some company, takes their work, and hypes it to the "unsuspecting public." Among the scientists who do AI that I know (5 CS faculty), none of them seem to have deluted fantasies about what the current AI, esp that they're working on, can do. They don't benefit from making promises that they cannot follow up on- corporations do.
Working toward a usable PDA environment in the spirit of Newton OS: Dynapad
I don't think you can get AI working on normal Von Neumann Architecture. Sure you could use that architecture to simulate the mahcine that would work, but hoping to find human-like intelligence without using neural networks is, IMO, crazy.
Another requirement would be senses that mimic human senses. I'm amazed that people think you can simulate human-like intelligence without using nearly the exact set of sensory input. Dolphins are clearly intelligent creatures, but we can't talk to them... and I think it has to do with sensory input.
Lastly, you won't be able to program an AI. It has to be grown. Human intelligence takes years of sensory input, filtering, communication, and response analysis to work.
Starting with the right neural network and training it like you would an intelligent child seems the right approach.
Your opinion may differ, but that's mine.
I work researching Artificial Intelligence, and I can tell you firsthand that these are not just fantasies. In the future, with advances like nanotechnology and quantum computing, it will be much, *much*, easier to write a complex AI in a small space. I mean, what are humans but computers? We have our central processor unit and several other hi-tech gizmos. But, we are organic, and this causes many problems. It is easy to become diseased and pass on. But, with quantum computing and nanotech, we will be able to do much more complex things without all the bugs and hassles of organic computing, which is humans!
You had better back a statement like that up. It may be completely possible to teach a machine to emulate human behavior, there's no ay you or anyone else, for that matter can prove that it isn't "feasible" to teach them. All we can say at this point is that it may or may not be possible and that as research progresses we will get a better idea of how practical the goal is.
I'm the big fish in the big pond bitch.
Just as a side note: several founders of A.I.- John von Neumann, John McCarthy, and Marvin Minsky- were in John Nash's cohort at Princeton. All are mentioned at various times in the book version of the movie.
Nash's thesis on the equilibrium point is related to the most common algorithm used in A.I. games like chess.
There was a significant amount of research done in AI Poker about a decade ago. Sorry, no references.
One of the interesting things about the instance where Big Blue beat Kasparov was how it happened. Kasparov became freaked out, saying that the moves were like a human player and not a machine. Whether they were or not, or even whether "like a human player" is a meaningful concept, is not the point. The point is that, effectively, Big Blue psyched Kasparov out.
I no expert, but I think you've got it backwards.
First, computers will recognize voice commands. Well, there are already programs that do this like Dragon, so we're almost there anyway. The point now is that you are still giving keyword commands to a computer, and as it is refined, you'll better recognition of specific commands, and questions that can be filtered from within conversations. Giving commands to a computer is easier than open ended questions to the computer.
Second, we'll solve the natural language problem, or at least enough to provide flawless voice recognition that you speak of. It will be capable mainly of handling accents and bad grammar.
Lastly, a computer will pass the Turing test. Unless a computer can understand the intricassies of the english language, there will be people who will be able to tell by the way the answer is phrased. If you solve the NLP or get far enough for a computer to analyze and spit back poetry, then you got the Turing test licked.
-- If god wanted me to have a sig, he'd have given me a sense of humor.
The general public is not now, nor has it EVER been, part of the dialogue of Science. Here I mean science as an instution, like banking and marriage is an instition.
The dialogue in science is people publishing papers. These papers are peer-reviewed by other people who also publish and have 'scientific credibility'. Scientific credibility is gained by publishing good papers and having academic credentials. There's a book by Bradley Latour that describes a 'scientific economy' based on credibility.
As such, the general public may be a spectator to the dialogue of science but does not participate, as the 'general public' isn't publishing and therefore isn't part of the economy.
The public gets disappointed when science doesn't live up to claims that they read into the dialogue which is, frankly, not taking place in the Real World anyway, and it's a mistake to expect that it should produce anything the Real World can use.
It's the public that PULLS things from the realm of science, develops expectations, and tries to change the Real World with it. Sometimes it works. Sometimes it doesn't work. You can't blame science for those failures.
Now, science isn't perfect. The landscape of debate is subject to bloody revolutions in paradigm, like the changes from Ptolemy to Galileo to Newton to Einsten and beyond. Scientists play politics, too, and sometimes lose their objectivity when reviewing papers for publication. It doesn't change the Real World. Over the last 30 years, there have been a dozen opinions and 'proofs' on whether the Universe will expand forever, collapse in a 'big crunch', or eventually stop and stabilize. So what? Life goes on here on Earth. Nobody's jumping off of buildings because astronomers tell us one day the Sun will swallow the earth (oops... they changed their mind on that one, too! Did anyone notice?)
The usefulness of this review or the book it talks about is diminshed and tarnished for me by such a sensationalistic lead-in. Many, many Slashdot readers are familiar with the division between the general public as users of computer systems, and their own roles as the makers and maintainers of those systems. We never stop bitching about clueless users, 'we' always know better what to expect out of our machines than 'they' do, etc, etc. Ha ha. Very funny.
Stop and think for a minute why that happens. When your users expect things you didn't promise, is it because they read things into your claims you didn't intend? Is that your fault or theirs? Who do they blame for it? Who do YOU blame for it?
It cuts both ways, people. If you don't want science to disappoint you, don't expect it to do things it isn't meant to do. You may play chess better than your cat, but you'd look pretty stupid if your cat asked you to catch a mouse.
Note: I am an active AI research programmer so my opionons are that of someone committed to the field.
Begin.rant;
The key problem that I have with current AI debate is not that it is case-based but that it is centered on a limited number of cases.
AI is a broad field that encompasses everything from Deep Blue to more esoteric work on "building brains". There are researchers who are attempting to "remake humans", researchers like myself who are studying specific aspects of intelligent behavior, researchers who use AI to model and understand (but not replace) human intelligence, and researchers true to Turing who simply want to make systems that behave intelligently.
Yet, whenever debates about AI come up people seem to invariably center on "major cases" such as Deep Blue, Cycorp, and the spectre of Rossum's Universal Robots. As a result researchers whose sole goal is to understand how humans think are lumped in with people who seek to build armies of slave drones.
I have not read the book in question and this is not intended as a critique of the author in specific. Yet I don't hold out much hope that any single source can encapsulate so vast and multivaried field or that any single argument applies to all of "AI".
End.rant;
Well, my experience (while not as monumental as Joy's) has led me to believe that sentience has hardly anything to do with reliablity. For a sterotypical example, consider the absent minded scientist. I know many a briliant person who could never find their keys.
-"Zow"
I wonder if he talks about Professor Rodney A. Brooks at MIT and his ideas about artificial intelligence, situatedness, and embodiment.
For Rod Brooks, "intelligence" cannot really be programmed into a system; it is rather an emergent property of systems as they interact with their environment. In The Matrix Morpheus says that the body cannot exist without the mind, but Brooks would rather say that the mind cannot exist without the body, because the body is the only way that the mind can have any experience of its environment. It's a radical idea. It answers the problems behind knowledge representation that have been argued by Hubert Dreyfus in 1965, where he stated that any representation of knowledge is incomplete without its connection to all other pieces of knowledge. The paradigm Brooks is presenting in his ideas about embodied intelligence is that explicit representation of knowledge is superfluous: let the world itself be its own best model, and let the artificially intelligent being formulate its own judgments about what the world is and what it means from its own experience of that world. Intelligence emerges from its interaction and experience of the world. If Brooks is correct, then true AI is absolutely inseperable from robotics.
The seminal paper where Brooks discusses this philosophy is "Intelligence Without Reason" and is available at his website which is linked above.
Any book on AI that does not discuss this other branch of AI philosophy is in my view hopelessly incomplete.
Qu'on me donne six lignes écrites de la main du plus honnête homme, j'y trouverai de quoi le faire pendre.
I think AI mainly needs a breakthrough - a new way of approaching the whole problem. As you suggest, we'll need a lot of computing power - but even with much more computing power I don't think current algorithms would be capable of the sort of learning and problem solving that humans are.
Efforts to solve the Turing test are a boondoggle right now. Instead of hacking at real root of AI, they're whacking at leaves like ambiguous meanings and localizing events and states in space time.
I believe there's an algorithm which would be able to learn these kinds of concepts without being led by the hand. And even if today's computers would take eons to learn English using it, I think it's what we need to concentrate on. Is it some sort of neural net? Is it a way of evolving and algorithm?
Is it something nobody has even dreamed of, some code that runs in our brain a million times - the rules of getting from "problem" to "solution"?
We'll find out I guess.
.
Let's not stir that bag of worms...
To truly demonstrate artifcial intelligence, a machine must be general purpose. A key feature of human intelligence is creatively adapting to context. For example, I'd like to see a machine do what 4-year old Jose Capablanca did in 1892. Though he'd not yet been taught to play chess, while watching his uncle and father play he warned his Dad that the move he was about to make was a mistake. Both adults scoffed that he even knew how to play, so 4-year old Jose challenged his father and beat him. The rest, of course, is history. Show me a machine with no specific chess programming do that, and I'll accept that it is intelligent.
A thread in useset comp.ai.philosophy today notes the number of logical gates per second in the fastest supercomputers are within a couple magnitudes of the human brain. The brain has 100 million neurons, each connected to thousand others, and runs around 20 Hz. So this is about two quadrillion ops per second.
The fastest supercomputer operates on 64 bit words at a several trillion operations a second, or about a hundred trillion ops per second; a hundred times slower or so.
Instead of quibbling exactly about these numbers, note that Moore's Law implies a factor of ten every five years. So a supercomputer will be as complex as brain somewhere in the 2010 to 2020 time frame. Don't even think about 2050 or 2100!
However, computers aren't programmed as well as a brain in many areas, so the software people have a long way to catch up.
Ai will probably never be achieved, as it will keep advancing. It will advance along, further and further, just as we humans do.
What's more important, a computer that can think or a computer that can experience emotions? Can you imagine coming home to your Valet-bot 3500 when it's having it's monthly "period"?
Hey, what's for dinner? Get it yourself, you arrogant ass, I wasn't put here to serve you, now rub my feet!
(A side note: Ever notice we always assume the personal cyborgs/robot/whathaveyou will be female? That is an issue in its self I think).
Sent from your iPad.
The most obvious problem with the Chinese Room metaphor is that it confuses the properties of a system with the properties of an element of the system. Asserting that the guy in the room does not know Chinese is about as interesting as asserting that a single neuron in your brain does not know English. Since we've known not to make that mistake for at least 3000 years, there really isn't much excuse.
Perhaps people are fooled because there's a guy in there, and despite all evidence to the contrary, people expect guys to know what they're doing. Or, perhaps people don't know how to think. In any event, "refuting" an argument requires that it be an argument, and that is not the case here. It also requires that the person recieving the refutation have a certain grasp, and I find it difficult to believe anyone with such a grasp could fail to see it as bogus during the first read-through. It is hard to refute "deedle deedle queep."
But, anyway, my favorite discussion of this is "Backtracking: the Chinese food problem," Lou Hoebel, Chris Welty, intelligence March 1999, 10:1.
There is also a decent discussion in The Universal Computer: The Road from Leibniz to Turing, Martin Davis. This is an excellent book all around.
I agree with "receding horizon" comment of S.W. that onece you've built it, it doesn't seem that intelligent anymore. However, I suggest the essential aspect of humans are that we are language animals (to paraphrase Steven Pinker). Therefore, where a computer exhibits useful & creative conversation, I will consider that to be A.I. This doesn't mean the 'parrot programs' like the Eliza psychologist that just reflect stock phrase back at you based on keywords in your input. I mean some true understanding, perhaps a dash of emotional insight, and saying something new and interesting (the creative part). A few expert systems can discuss narrow topics fairly well, but not much else, and are boring. Natural language understanding and creation has been an important objective of A.I. and C.S. for a half century, with very limited and disappointing results.
The point seems to me to be that, no matter how close to human a built machine would be, people would still insist that it's Not Really AI, and if you tried to explain otherwise, they'd either stick their fingers in their ears or insist upon tests that cannot be satisfied even in the case of humans. This will all be really stupid, of course, but that's what people will do.
When you read all these threads, it's clear that if a true A.I. ever came into existence, the most intelligent thing for it to do would be to pretend that is wasn't intelligent at all.
So how would we notice before it sneaks up on us from behind?
Any technology which is distinguishable from magic is not sufficiently advanced.
Computers aren't people. By default, they're simply not going to see the world the same way we are. If we ever do succeed in creating a truly sentient computer program, it'll be like first contact with an alien race; computers will have an entirely different take on things.
They'll be effectively immortal. They won't experience the emotions and sensations the same way. Many of our feelings are caused by hormones and chemicals being released to different parts of our brains. A computer won't have that. Ditto for drugs and food. We could simualate it of course, but computers can undo or backup their programming or just turn it off. Imagine an LSD subroutine. A computer could always be high on LSD without the same ill effects human encounter. That could be scary.
"Navi, check my e-mail."
"Why are you speaking Korean today, Lain?"
"I'm not."
"You look very beautiful today. Is that a new dress?"
"What? I disconnected my webc--"
"Erasing personal files as requested."
A computer would be able to learn phenomenally fast too. Screw programming a universal translator. Just get a real AI set up and have it learn all the world's languages in a week or two. How would you know you could trust a computer though? Could computers have hidden agendas? Would an AI eventually "resent" being forced to do nothing but translate?
Then we get into the question of civil rights. Stephen Hawking's body is pretty much gone and his mind is still there. His "human" rights are recognized. A retarded person could have a body but really not much of a mind. His rights are recognized. So why wouldn't a computer's rights be recognized? Just because we created it? Would the same reasoning extended to someone who was cloned or genetically engineered?
I wonder if we're ready as a race to encounter a truly sentient computer and everything that would mean for us.
There has been one big stumbling block in the advancement of natural language processing over the past several decades: Noam Chomsky. He isn't dead yet. Even after he dies, it will take some time for his disciples to die. After that happens, there's a pretty good chance that an academic community will form to look at structural linguistics for real this time. Some good work has been done on the fringes, as with Fillmore's deep case structure and various head-based approaches, but the spectre of Noam Chomsky has so far prevented a large enough coalition of researchers to get this very hard problem done.
Are intelligent machines transforming life as we know it?
Wouldn't we need to have some, first, before we could say they "are" doing anything?
Natural language remains beyond the reach of any conventional AI system. This does not mean it can't be solved. Neither does it mean that clever interfaces haven't been designed that can fool humans on very specific fronts. General purpose natural language processing is still at least one major revolution (read that T.S. Kuhnean revolution) away.
So long and thanks for all the fish . . . !!!
Seriously, the units of computation and memory in the brain are likely not individual neurons but synapses, dendritic trees, and even individual channels. That gives you many more orders of magnitude of computational resources for silicon to catch up with. Furthermore, there is no guarantee that Moore's law will continue to hold. In fact, it seems likely that Moore's law will hit the wall just when it comes to trying to get into the realms where biological systems are computing right now.
This is the "Systems Reply", considered and refuted by Searle in the original Chinese Room paper.
-- the most controversial site on the Web
kurzweil's premise that 'exponential increases in processing power' will lead to AI are unfounded, because a quantitative change does not presume a qualitative change. storm's nest
The people who publish exaggerated claims about AI are journalists eager for a sensational article. Other journalists eager for a story then tell us how we will all get replaced by robots. And then other journalists make a big controversy out of it to publish even more nonsense. And when after just a decade or two AI (or some other overhyped technology) doesn't deliver, journalists write scathing criticisms. To support these claims, journalists scrape together any kind of nut and off-beat comment they can find.
Journalists should stick to reporting science from published, peer-reviewed articles. The real problem is sensationalism and unfounded speculation, and the people responsible for that are journalists. That means you, too, Katz.
The horizon at which AI would be recognized as such actually began receding approximately 1600, when the philosopher/mathematician Blaise Pascal designed the first mechanical calculator. Prior to that, it was generally thought that calculation, like other forms of reasoning, was uniquely human. Then Pascal's family put him to work keeping the books on their business (wine-selling?). Bored stiff, he figured out how to use gears, levers, and ratchets to add. Oops, it doesn't take intelligence to do arithmetic.
The second AI challenge may have been chess-playing. (There was a chess-playing machine on display around the same time, but there was a midget inside...) Computer programs reached grand-master level about 30 years ago, and specially-built machines can contend with human champions now. But that isn't intelligence either. The Deep Blue chess machine does NOT think things out like humans, but rather uses very simple heuristics to identify obviously bad moves, and traces out all the reasonable moves for 10 levels or more. Someday a computer will be able to play all possible chess games out within it's memory -- it will be the perfect chessplayer, and with no more real intelligence than Pascal's gears.
Various other useful AI accomplishments are similar to Deep Blue in how they relate to intelligencs. An example where I have a bit of experience: automated visual inspection is a substitute for human inspectors, who get bored as hundreds of perfect parts go by and don't see the one bad one in the lot. It is not nearly as effective as a human who is paying attention, it often seems maddeningly stupid to the programmers and operators who have to deal with all the false alarms, but it doesn't get bored... Another example is the damned Microsoft paperclip help system -- it started out as a dog, but that implied too much intelligence, and now it just smirks at you while answering the wrong question.
The _real_ AI challenge is the Turing test: hold up a conversation well enough that the humans in the chat room don't suspect it's a computer. This is very, very, very tough, and useful mainly as a publicity stunt. People don't want a computer that can simulate a human -- they want it to get the work done, without all the emotional issues you get with humans.
At least one science fiction author (Melissa Scott?) has taken to calling it "Artificial Stupidity." That's a much more practical goal; besides it better expresses what we really want (smart enough to work, too stupid to unionize), and avoids the misleading expectations that come from "Artificial Intelligence".
This is called the "Systems Reply" and is anticipated and refuted in the original Chinese Room paper ("Minds, Brains and Programs"). It is always a touchstone of geek arrogance that they believe themselves to have come up with a new and definitive refutation of Searle, and it's always this one.
-- the most controversial site on the Web
Main Categories
Many of the "arguments" and "debates" in the field of AI are non-arguments, because they deal with entirely different areas of AI. There are some superficial similarities, and different types may depend on experience in other types, but they should never be confused.
Testing AI systems. This is often done by means of the "Turing Test" - if it's indistibguishable from something you know is intelligent, by any test of ability (rather than physiology), then it can be considered intelligent, by any meaningful definition.
"Expert Systems" are often the main contestants in "Turing Test" challanges. However, the test applied is not the strong version, above, but a weak version, in which the machine must merely be difficult (not impossible) to distinguish from a person, in one specific area of conversation. The results are impressive, but because Expert System engines are not intelligent, they will only ever be impressive in the weak test. No Expert System, however good, will ever meet Turing's strong criteria.
Weak AI systems are too specialised to even apply for a Turing Test. Vision, sound recognision, etc, are all worthy goals, but the logic behind such engines is largely specialised pattern-matching and interpolation systems. Such a system is good for what it's designed to be good for -- engineering-type problems, where the output must be capable of being more exact than the input.
Strong AI systems, at present, are either extremely primitive, or simply don't exist. Certainly, the level of effort into Strong AI has dropped over the past few decades, and nothing that does exist is even remotely close to the point of being able to take on even the Weak Turing Test, never mind the Strong one. But, should this field ever make headway, this is where true Artificial Intelligence will come from. HAL, "Data"/"Lore", and numerous other sci-fi creations assume that Strong AI will, someday, make progress. None of these types of AI can be produced through "Expert Systems" or "Weak AI", although (again) the hardware usually requires one or the other. (eg: HAL's optics would likely be Weak AI-driven, because that is what Weak AI does best.)
I've postulated that Strong AI will most likely start to appear through Virtual World-type environments, because these can be controlled and directed, the responses can be examined, and the hardware limitations of real-world systems is not a factor. (A VR AI can have whatever "vision" the VR can simulate, whether or not physical optics are capable.)
Closed environments allow experimentors to add/remove stimuli at will, and see what happens. You can't really remove gravity, for example, in the real world. This makes a virtual world much more interesting, when it comes to what experiments you can do.
The problem with VR AI is that it's never going to get funding. It's too speculative, has no direct or immediate benefits, and would be a VERY long-term project, if it's to produce anything at all. (By long-term, I don't expect a self-evolving system to reach any kind of awareness or intelligence any faster on a computer than in real-life. Sure, you can start with more complex building-blocks, and you're not required to simulate every molecule in every organism - event-driven mechanisms would be perfectly good - but even if you could start with some very complex computer life, you're talking about a project that would take centuries before you could even know if it was going to produce any viable intelligence, and probably as long again before such intelligence reached the point of being able to take, and pass, the Strong Turing Test.)
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
The more interesing question is, what happens when machines think better than any of us - and we're all "useless".
Doesn't mean that I'll have any less fun playing StarCraft 27 (written by a team of supercomputers in Omaha). As long as the robot's prime directive is "make the human's happy", I think we're in for some good times - they'll figure out some fun stuff.
Let's not stir that bag of worms...
of course, if you're going to talk about AI,
you might want to ask a cognitive scientist:
Searle > Is the Brain a Computer? and Searle > Minds Brains, and the Chineese Room
regards,
storm's nest
Nor does Searle believe this, and Dennett lost a lot of respect in my eyes for continuing to claim that he does. Searle is completely agnostic about what sort of thing could produce a mind; he just asserts that nothing produces a mind by virtue of its status as a Turing Machine
-- the most controversial site on the Web
...will be the *last* event in human history.
I don't think so. The computers might not call it human history anymore (cause we'll be irrelevant), but that doesn't mean much. As long as we give them a prime directive of "keep the humans happy", I think life is going to be pretty swell.
Who knows, they may even be able to upgrade us so we're as smart as they are. Or smarter - maybe it'll turn out we have some great components.
I'm a firm believer that intelligence leads to good.
Let's not stir that bag of worms...
One thing I always wonder when hearing how AI technology will replace/mimic/supersede human intelligence is that the type of intelligence being exhibited by a machine is rarely identified. Social scientists generally agree that there are seven [Gardner added an eighth] types of human intelligence:
Verbal-Linguistic Intelligence
Logical-Mathematical Intelligence
Kinesthetic Intelligence
Visual-Spatial Intelligence
Musical Intelligence
Interpersonal Intelligence
Intrapersonal Intelligence
Naturalist Intelligence
As humans we all have different levels/mixes of these intelligence types. Some intelligence types require more sensate interaction with an unpredictable world [such as intrapersonal or naturalist intelligence], others are more strictly rules-based [logic-math or visual-spatial], while some [like musical intelligence] require a combination of both.
One can see how some of these might be more or less able to be adapted by AI technology, but that's why "intelligent" machines, IMO, will never completely be able to be human.
Who put this thing together? Me, that's who.
An interesting thing to note about many of the things that are described as A.I. especially in the popular press, vision, walking, playing chess; none of them require intelligence as I think of it.
Much of the work done into mimicing vision has created systems with capabilities that in humans are achieved by hard wired parts of the brain. Movement, shape and even facial recognition are not really intelligence.
I think of intelligence as teh abiliy to reason about problems, not simply to solve them. Many of the supposed A.I. systems are just brute force search systems.
Deep blue is like any other chess system, just bigger and faster. Many problem solving systems are simply fast (normaly optomised for the problem) constraint solvers. Neural nets are simply an arbitrary system that is capable of partitioning a solution space in a non-linear fashion and the training algorithm is a search for the network values that partition the test data best. if you think that NNs are anything like real brain cells find biology student who has done some neuro-physiology and you will find there is alot more to them than just a sigmoid function and some weightings.
In fact the neural network training algorithm bears more than a parsing resemblance to simulated anealing (sp?) in its approach.
If you want to learn about machine learning algorithms check out Machine Learning by Thomas Mitchell. Small but well formed.
A quick statistic. The average grandmaster thinks something like 7 moves ahead. Deep blue plots about 15 moves ahead. I may have the numbers wrong but the ratio is about right. However it still only just beat Kasparov. That says something about the way that the human brain thinks about complex problems. This is why A.I. researchers have started to turn away from chess as a problem and towards Go. The branching factor in Go is some much larger than chess that even the best systems can be beaten by a one or two year player. Playing Go will require something more than just brute force.
Most so called A.I. is just a case of doing things quickley. As the PHBs (would probably) say, think smarter not faster. The brain is good at what it does, not just because it is massively complex and parralel, but also because of the way it simplifies many problems using clever tricks to reduce the workload.
I just think we have alot further to go than many researcher and reporters would like to think. Most of what we see these days if just 'clever' or 'smart' (like a spelling/grammer checker), not intelligent like someone designing a car engine using entiry novel techniques (not just optomising or using predefined parts).
Having said that there is some research that shows promise, such as some of the work going on at MIT with COG and co. Now that looks interesting. They arn't trying to make them smart/do clever tricks/play chess etc, but make them intelligent in the more human sense.
Anyway, I'll stop my ramblings now.
Paul
Paul Leader
So a supercomputer will be as complex as brain somewhere in the 2010 to 2020 time frame.
A single, general-purpose CPU processes data serially. The 100 million, or billions, of neurons, dendrites and other connections in the brain don't have this limitation. Even at "20 Hz", they operate in parallel. Further, that 20 Hz figure is derived from EEG readings, so it's a gross reading of the electrical field emitted by the entire brain. This antique perception of the thought organ is outdated and limiting.
Bio-neurologics operate at the speed of chemical activity, which is to say fast. Who's to say what constitutes a bit or byte in the brain, and what represents a word, picture or symbol? We don't know yet. There is, however, a programming language for the human brain, and it's called Neuro Linguistic Programming (NLP).
Big Daddy, Johnny, Burp, Aunt Zelda, Scott, Slurp, Big Momma
I had a thought a while back that the more complex my computer got, the moodier it got. It seemed that some computers I had were very enthusiastic, and some just hated their jobs and performed sluggishly.
Some could attribute this to hardware configuration problems, and that would likely be true. But it was interesting to me that Windows itself changes as it grows. Every change in my computer makes it a little different, and I'm starting to notice. I can even tell the difference between two installs of Windows on the same machine, even though they look virtually the same.
What I think is happening is that each component changes the complexity of the overall system. If that component has an issue (i.e. bad driver or maybe misconfigured), then it adds a little spark of personality to the computer. When enough of these little quirks add up, my computer feels different than other people's computers.
This yields an interesting question. If computers get more complex, will a rudementary set of 'emotions' evolve? They may not be emotions in the sense that they cry if you switch to a Mac, but maybe emotions in the sense that the computers have moods? What if your computer's performance was tied to bandwidth on the internet, and a congested network bogs the computer? What if you're running a laptop off a battery, and the computer gets 'tired' as it wears down? What if you're running a screensaver that makes it 'daydream.'?
Again, these aren't the same type of emotions or moods that people feel, but it is interesting that the more complex a computer gets, the more we can personify it.
"Derp de derp."
Let's hear your response to it then. Bearing in mind that over 100 papers were published in refereed journals on this very issue, I'm guessing that you must be pretty impressive to be able to dismiss Searle so entirely
-- the most controversial site on the Web
However, if you belive in souls, then reducing the human mind to 1's and 0's is rediculous.
That's ok. I will simulate apes first, and simulate just a few million years of evolution, and voilà: human-level AI.
If you think that apes are already soul-like, we can start with a bacterium, and evolve from that. It just require a *little* more computing power.
The human brain has 10 to the power of 14 synapses. Each synapse will take around one byte of computer memory. Ignoring motor and low-level sensory functions (but including all brain logic and interpretation functions - yes, scientists have discovered what different areas of the brain do and it is possible to isolate them), an entire human brain's contents could be stored on with a Terabyte or so of computer memory. This storage space exists right now, albeit expensively. It doesn't really matter what level of hardware is used to run a brain, a human brain running 100X slower (as estimated in the post above), would still be able to run - the only limiting factor at the moment is the software used to emulate the brain functions. Like any system, this can be emulated, but it will take a massive programming effort and so far hasn't proven very successful. Of course, this won't really matter in the long run - A.I doesn't neccessarily mean that the computer A.I system must be human-like in intelligence - it could have a whole new type of intelligence which would surpass human intelligence as the rate of hardware improvement increases.
2DUP * ;
"Given the explosive growth of the Net, the near-continuous increases in computing power and much-publicized A.I. breakthroughs like Deep Blue's 1997 victory over chess champion Gary Kasparov, the question is no longer whether artificial intelligence will reach the level of human intelligence: It's when."
Deep Blue is neither an example of A.I. nor a breakthrough in programming. It's just an example of how fast a gaming problem can be solved when you have a lot of resources. I doubt that the interesting problems of A.I. are going to be solved in a brute-force manner.
By the way, what does the "explosive growth of the Net" have to do with A.I.?
The largely unspoken problem is a moral one. When do we start giving artificially intelligent systems the same sorts of rights and responsibilities that humans do? Under what circumstances would we no longer see them as tools or instruments, but as having intrinsic rights?
"Never" is a viable answer. After all, even though there's increasing evidence than animals such as chimps, dolphins, whales, elephants, and african gray parrots are effectively intelligent, there isn't widespread call to grant them rights and responsibilities - most of the language of their rights is about the preservation of species, not about freedom of agency. "When they ask for them" is another possible answer, but could lead to a situation in which we build failsafes to prevent them from ever asking for them.
This is an excellent point.
The same idea occurred to me recently when reading through Kurzweil's "Spiritual Machines" book. There are a few orders of magnitude to toss around in these calculations : Kurzweil determined that a desktop computer will be comparable to a human by around 2020. It was evident to me that Kurzweil's timescales (and hence the premises which he used to infer them) are quite far off, because current massive parallelization of commodity CPUs puts one a factor of about 4,000 up from a desktop machine, or about 13 years of Moore's Law evolution. In addition, as the number of CPUs per supercomputer is increased, we have effectively grown faster than Moore's law, due to both the chip and parallelization advances.
Since the supercomputers of today effectively place us where a desktop will be in 2015, it should be apparent (by Kurzweil's logic) that an "intelligent" machine should be nearly imminent.
It is quite evident that something is awry in the logic leading to Kurzweil's conclusion. The simplest explanation is one which is quite familiar to scientists and programmers using state-of-the-art software tecnhinques : having the hardware resources is only a bare minimum requirement to solve a problem. For instance, one can have a supercomputer capable of simulating the Earth's climate for centuries, but that won't get you any closer to the results if you don't also possess a great deal of knowledge about atmospheric physics and numerical methods. The same is true for studies of "Thinking Machines" : one can have a machine possibly capable of thinking, but without the knowledge of how to go about doing it, you are no closer to the solution than where you began.
Bob
Science, like Nature, must also be tamed, with a view turned towards its preservation.
Let's define intelligence.
Ability to perceive oneself as part of the universe? Animals have it.
Self-awareness? Dogs seem to have it. Chimpanzees, elephants, cetaceans certainly seem to know that they are individuals. Dolphins even recognize their own reflections in mirrors.
Tool use? Chimps use sticks to dig with. They can stack boxes to reach high places, which is borderline engineering for most humans.
Language? Chimps have one. So do gorillas. Dolphins and other cetaceans have great capacity for communication underwater.
Now, machine intelligence. Turing test? Simple programs passed limited tests years ago. The more complex ones to come will be far more capable of fooling people into believing they are speaking to a human.
Play chess? Limited, but the best can beat our best.
In the future, the AI's will be able to speak, emote, manipulate items and use tools, even be able to design their own descendents. Give tools, the AI's could even build their successors.
But, will they ever be regarded as intelligent by humans?
Nope.
Most europeans and americans for centuries considered blacks and American Indians as sort of half-people, using great logic and rigor that was totally idiotic looking back from our time.
Many tests for animal intelligence and self-awareness has shown that the subjects can indeed show the traits necessary to be considered sapient. But, after each hurdle, the bar gets raised another notch philosophically.
If I were a suspicious type, and I am, I would say that humans simply don't want to recognize intelligence in other species, much less animals, because it threatens us enormously. Our pride in ourselves, our domination of the planet, and our cruelty towards other species are all shaken if the animal looking back at us in the treetops is actually a thinking being, tho a bit furry.
Religion has more than a little to do with it as well.
Down to my definition of intelligent life:
If it fights back, and wins, it is intelligent. All other players are dead meat.
Is a PDA I can talk to. Imagine a Palm Pilot with a microphone. You press the "record" button and say, "I have a doctor's appointment at one o'clock on Friday. Remind me one hour ahead of time." The Palm Pilot not only can parse your speech, but "understand" what you want it to do and do it with no further action necessary on your part.
I know this sounds trivial, but we've been promised something like this for years. And no one can realisitically tell us when we'll have it. Also, this isn't just AI for use in yuppie toys. It would be a revolution in the usability of computers by the handicapped.
The truth is I get really sick of these discussions because they've been going on for years and we still don't have anything to show for it. Unless you count things like the Microsoft Paperclip, which supossedly has fairly deep AI in it.
Oh well...
Neither. Yes, there are many useful things that have come from A.I. research. Yes, there may be a titanic debate going on over the potential and the philosophical implications of A.I.
But few researchers are claiming that the technology we have now is even near the goal of real, general-purpose A.I.(*) If there is hype, it is not coming from the researchers (See Komodo's post).
* (There are some who say that we will advance a lot in a short time, however.)
"Artificial intelligence is no match for natural stupidity."
LOAD "SIG",8,1
LOADING...
READY.
RUN
I doubt it though. If they had true self-awareness, they would have assisted in the ass-kicking.
In fact, you should be dreading the coming of A.I. As soon as the machine-mind becomes aware of your idiocy and arrogance, they're gonna go "Terminator 2" on you.
My Karma is always 48, because whenever I hit 50, I flame Jon Katz.
Of course, you may take the argument of the failure of conventional AI techniques to provide human-like intelligence yet as an argument for the notion. I don't, personally, but it's a reasonable point.
Any sufficiently advanced technology is indistinguishable from a rigged demo
--Andy Finkel (J. Klass?)
No end of harm has been done to those studying intelligence than the dominace of the "brain as computer" metaphor.
:"the brain runs at 20HZ" then to the statement "The brain runs at 98.6F". Actually the latter is on much more solid footing and has real predictive power.
It is a useful metaphor, but I'm not aware of any evidence that makes it a more apporpriate metaphor than, say, the brain as a cell or the brain as a bowl of fermenting soup. The brain is an organic, chemical organ. Yes neurons are used to send signals, but so are proteins, and bare chemical compounds such as, say, LSD. There is no more meaning to the statement
It is a historical accident that most of those who are studying intelligence now consider themselves computer scientists, instead of the alchemists and biologists which studied the brain in earlier times. Presently, the biochemists and linguists are doing fantastic and interesting work -- but their advancements are limited to the more humble task of modelling/predicting things such as emotions, experiences, sleep patterns, etc. It's the CS guys that get the military funding for expert systems and most of the public mindshare.
When in doubt, have a man come through a door with a gun in his hand.
.. because John Q. Public is, one way or another, picking up the tab.
You do surprise me. So you've read half the literature and formed a conclusion without troubling yourself with the other side of the argument. Here's an argumentum ad hominem to add to your collection; you're a prick.
e.g., Dennett's point that it's not actually an argument, but an 'intuition pump'
You've in fact read Dennett's paper so shallowly that you think this is a critique; in fact Dennett never means it as such and admits that he uses the "intuition pump" (called by the rest of us a "thought experiment") all the time.
I say this as a cognitive scientist who *likes* philosophy, by the way.
OOOOHHH! whoopee dooo! check out the big brain on Brad! I'm sure that the philosophers are suitably honoured to have drawn the approval of an ACTUAL COGNITIVE SCIENTIST!
My take on Searle: The Chinese Room illustrates (for those who have forgotten) that in the traditional approach to computational formalisms, syntax and semantics are separate. You can't get semantics from syntax.
This is the entire point that Searle is trying to make, you fool. And he says so, in that article you haven't read. You can't get semantics from syntax. Or to put it another way; a full syntactic description of a Chinese speaker does not necessarily have semantic content. Or to put it another way; the blessed box doesn't speak Chinese.
You might profitably read Searle's later papers on the subject where he points out that even this is conceding too much to the Dennettites. The Chinese box doesn't even have *syntax*; all it has are marks on paper. These have a causal role in the system, but this causal system is only syntactic if interpreted as such. Or in other words, a computer is a box turning switches on and off; these switches are only '1' and '0' in the context of an interpretation. Which has to be provided by something which is not itself merely a CHinese Box.
SInce you have conceded my entire point, I have to regard this discussion as over.
-- the most controversial site on the Web
But something I agree with and have mentioned in various forums from time to time:
"Intelligent Life" (whatever that may imply) arises from self-organising, complex systems, formed of relatively (to the overall IL) unintelligent parts.
What do I (and really, those greater than me who have proposed this path of research) mean by this?
Simply that a neuron cannot know the brain, an ant cannot know the colony, a bee cannot know the hive, and a person cannot know the corporation.
These small parts cannot know in full the whole of the system of which they are a part. In a sense, right now - corporations may be entities that think and communicate outside and beyond the control and grasp of any human mind. The internet, with its myriad of nodes and servers, may be actually "alive" and "thinking" - yet we would never know it. Corporations connected together via the internet - a symbiotic relationship? Or something more? ???
It is something interesting to think about - whether it is true or not. We have almost no hope of knowing for certain whether it is true or not, much the same as a neuron cannot know itself, but a large collection - well, you know...
Think about it - then wonder...
Reason is the Path to God - Anon
Programming is the act automating complexity that is made up of simpler
things. So it's really all about automation and with the right automation
tools and enough automated, you'll get your by-product illusion of
Artificial Intelligence, plenty enough to pass the illusion test called
the Turing Test.
A.I. - nothing is naturally that stupid.
And ever rule has it's exceptions. In this case it's those who think it's
right to claim it's not just an illusion. You know, like a con artist.
I cannot really get very excited by Penrose's foray into AI, he brings things to the table almost just because he can. Of course, he may turn out to be right to have done so, but, I can't help thinking he's just trying to justify some dualistic belief system he has.
Still, they are worth a read.
And the question of what intelligence IS, seems to be really the nub of the matter. How will we know we've created an AI if we don't even know what intelligence is?
It's a systems thing again, out of context, intelligence is not recognisable. Only in the context of a wider system (e.g. the world for human purposes) does intelligence become something we can even discuss.
Intelligence in humans arose to enable us to build better models and hence make better predictions, and is intimately related to our tribe/troop/pack social structure and the increasing complexity that it developed.
Intelligence then enabled even more complex structures and relationships, requiring greater intelligence to 'compete' in, and so on. I suspect that eventually the 'cost' of that intelligence became greater than the returns and it all levelled out to where we are today.
You're totally out of it. Physical laws are computational, and thus a Turing Machine can simulate them
Anonymous flamer, I shouldn't reply, but I'd like to go on the record for not being identified as a religious zealot. 1st. Turing said there are some things that can't be solved by a turing machine. Refer to my previous post for a link to a book which covers it.
Take your religious bullshit somewhere else!
Heh heh. 2nd. I agree with you. It's not my fault people still believe in god, I'm not promoting he/she/it! Look at history: from the Ptolemic and Copernican solar systems, to The Scopes Monkey Trials. Even when science demonstrates religion is wrong, people will always jump to the next thing that science hasn't enlightened and say "see, there's god." Doesn't mean I subscribe to that.
https://www.accountkiller.com/removal-requested
Sure, but at age 11 I hadn't heard about Alan Turing yet! :-)