Slashdot Mirror
Where's HAL 9000?
An anonymous reader writes "With entrants to this year's Loebner Prize, the annual Turing Test designed to identify a thinking machine, demonstrating that chatbots are still a long way from passing as convincing humans, this article asks: what happened to the quest to develop a strong AI? 'The problem Loebner has is that computer scientists in universities and large tech firms, the people with the skills and resources best-suited to building a machine capable of acting like a human, are generally not focused on passing the Turing Test. ... And while passing the Turing Test would be a landmark achievement in the field of AI, the test’s focus on having the computer have to fool a human is a distraction. Prominent AI researchers, like Google’s head of R&D Peter Norvig, have compared the Turing Test’s requirement that a machine fools a judge into thinking they are talking to a human as akin to demanding an aircraft maker constructs a plane that is indistinguishable from a bird."
He talks mostly in this article about how the focus has been on developing specialized software for solving specific problems and with specialized goals, rather than focusing on general AI. And it's true that this is part of what is holding general AI back. But there is also something that Loebner is perhaps loathe to discuss, and that's the underlying (and often unspoken) matter of the *fear* of AI.
For every utopian vision in science fiction and pop culture of a future where AI is our pal, helping us out and making our lives more leisurely, there is another dystopian counter-vision of a future where AI becomes the enemy of humans, making our lives into a nightmare. A vision of a future where AI equals, and then inevitably surpasses, human intelligence touches a very deep nerve in the human psyche. Human fear of being made obsolete by technology has a long history. And more recently, the fear of having technology become even a direct *enemy* has become more and more prevalent--from the aforementioned HAL 9000 to Skynet. There is a real dystopian counter-vision to Loebner's utopianism.
People aren't just indifferent or uninterested in AI. I think there is a part of us, maybe not even part of us that we're always conscious of, that's very scared of it.
What political party do you join when you don't like Bible-thumpers *or* hippies?
Not AI enough? - http://www-03.ibm.com/innovation/us/watson/index.html
Join the Slashcott! Feb 10 thru Feb 17!
Too many decades of lofty promises that never materialized has turned "AI research" into a dirty word...
Palm trees and 8
Forget HAL, where is Cherry 2000!
Strong AI has always been the stuff of sci-fi. Not because it's impossible, but because it's impractically difficult. We can barely model how a single protein folds, with a world wide network of computers. Does anyone seriously expect that we can model intelligence with similar resources?
Evolution has been working on us for millions of years. It will probably take us hundreds or thousands before we get strong AI.
Give me Classic Slashdot or give me death!
"The question of whether a computer can think is no more interesting than the question of whether a submarine can swim."
Non-Linux Penguins ?
I would argue that placing emphasis only on the Turing test itself is a distraction from the broad field of AI. For example, there is a ton of really cool work coming from various labs ( http://www.ias.informatik.tu-darmstadt.de/ , http://www.cs.berkeley.edu/~pabbeel/video_highlights.html).
There are many achievements met and progress made, e.g. Peters group's ping pong robot, just not the ones researchers promised many years ago.
I tend to think we need to split out "Artificial Sentience" from "Artificial Intelligence." Technologies used for expert systems are clearly a form of subject-matter artificial intelligence, but they are not creative nor are they designed to learn about and explore new subject materials.
Artificial Sentience, on the other hand, would necessarily incorporate learning, postulation, and exploration of entirely new ideas or "insights." I firmly believe that in order to hold a believable conversation, a machine needs sentience, not just intelligence. Being able to come to a logical conclusion or to analyze sentence structures and verbiage into models of "thought" are only a first step -- the intelligence part.
Only when a machine can come up with and hold a conversation on new topics, while being able to tie the discussion history back to earlier statements so that the whole conversation "holds together" will be able to "fool" people. Because at that point, it won't be "fooling" anyone -- it will actually be thinking.
I do not fail; I succeed at finding out what does not work.
Back in the early 1950s, it was thought that the real prize of AI was to get a computer able to beat the best human chess player consistently. The reasoning at the time was that the only way this would be possible was for breakthroughs to happen in AI where a computer could learn to think and could reason better at chess than a human. Fast forward to 10 or so years ago where IBM realized that just by throwing money at the problem they could get a computer to play chess by brute force and beat the human champion more often than not. So I'm not surprised that some AI people discount the Turing test. I am not an expert in the field but it seems to me that AI is a heck of a lot harder than anybody realized in the 1950s and we may still be decades or even centuries away from the kind of AI that people 60 or so years ago thought we'd have by now. Part of me does wonder if maybe just like how AI research in chess took the easy way out by resorting to brute force that now it's they'll just say the Turing test is not valid rather than actually try to achieve it because to pass it would require breakthroughs nobody has thought of yet and that's hard.
He talks mostly in this article about how the focus has been on developing specialized software for solving specific problems and with specialized goals, rather than focusing on general AI. And it's true that this is part of what is holding general AI back.
No, that's not true ... that's not at all what is holding "general AI" back. What's holding "general AI" back is that there is no way at all to implement it. Specialized AI is actually moving forward the only way we know how with actual results. Without further research in specialized AI, we would constantly get no closer to "generalized AI" and I keep using quotes around that because it's such a complete misnomer and holy grail that we aren't going to see it any time soon.
When I studied this stuff there were two hot approaches. One was logic engines and expert systems that could be generalized to the point of encompassing all knowledge. Yeah, good luck with that. How does one codify creativity? The other approach was to model neurons in software and then someday when we have a strong enough computers, they will just emulate brains and become a generalized thinking AI. Again, the further we delved into neurons the more we realized how wrong our basic assumptions were -- let alone the infeasibility to emulating the cascading currents across them.
"General AI" is holding itself back in the same way that "there is no such thing as a free lunch" is holding back our free energy dreams.
But there is also something that Loebner is perhaps loathe to discuss, and that's the underlying (and often unspoken) matter of the *fear* of AI.
We're so far from that, it humors to me to hear questions and any semi-serious question regarding it. It is not the malice of an AI system you should fear, it is the manifestation of the incompetence of the people who developed it that results in an error (like sounding an alarm because a sensor misfired and responding by launching all nuclear weapons since that what you perceive your enemy to have just done) that should be feared!
People aren't just indifferent or uninterested in AI. I think there is a part of us, maybe not even part of us that we're always conscious of, that's very scared of it.
People are obsessed by the philosophical and financial prospects of an intelligent computer system but nobody's telling me how to implement it -- that's just hand waving so they can get to the interesting stuff. Right now, rule based systems, heuristics, statistics, Bayes' Theorem, Support Vector Machines, etc will get you far further than any system that is just supposed to "learn" any new environment. All successful AI to this point has been built with the entire environment in mind during construction.
My work here is dung.
Artificial Intelligence is just that Artificial. Big Blue has zero actual Intelligence, but has plenty of ways of accomplishing a task (chess) that usually requires actual Intelligence. The article has confused Machine Intelligence and Machine Learning with Artificial Intelligence. The problem is that in those areas no one is "best suited". If we knew what we needed to do for Machine Intelligence to work then we'd have a Hal 9000 by now. Instead we have Watson, though impressive, is a long way away from Hal.
Festo's Smartbird is hardly indistinguishable from a real bird, but it is much more so than say da Vinci's ornithopter. A slow and steady progress can be charted from the former to the latter. At some point in the future, the technology will be nearly indistinguishable from a real bird, thus passing the "Norvig Test".
That's the whole point of the Turing Test; it's supposed to be hard and maybe even impossible. It doesn't test whether current AI is useful, it tests if AI is indistinguishable from a human. That's a pinnacle moment, and one that bestows great benefits as well as serious implications.
Personally, I think it will happen; maybe not for 50, 100, 500 years...but it will happen.
I'm sorry, but your opinion seems to be wrong.
I'm not. AI is to real intelligence what margarine is to butter - it's artificial. It isn't real. You're never going to get a Turing computer to actually think, although some future chemical or something machine may.
However, you could get to the point where intelligence was simulated well enough that it appeard to be sentient.
Which leads to what I fear, that people like those in PETA will start a "machine rights" movement, where it may be illegal for me to shut off a machine I built myself!
Luckily, I'm not likely to live long enough to see it. Some of you might, though.
Free Martian Whores!
Im afraid Apple wont let me do that, Dave.
If a computer could think for itself, and solve problems on its own, it would logically conclude the fate of humans in less than a second. Unless we could confine that intelligence so it can't access the Internet, than those who posses the technology would rule the world. Either way, super intelligence is bad for humans.
-- By all means let's be open-minded, but not so open-minded that our brains drop out.
It's asking for the world's best stage magician to create real hovering women.
"If you REALLY fool me, it will be true!"
Nonsense.
"Flyin' in just a sweet place,
Never been known to fail..."
it's artificial. It isn't real. You're never going to get a Turing computer to actually think
Why not? We evolved into sentient beings from non-sentient organic matter, why couldn't the same thing be possible with silicon based intelligence?
^^vv<><>BA
She gets all huffy when you ask her that.
Good-bye
Thats is why we seek out each other and other intelligences in the universe. Steven Pinker captured the gist in calling it The Language Instinct. Humans go more or less crazy in perpetual, involuntary solitude.
A computer intelligence is probably the best long term prospect for an interesting intelligence to communicate with. We've been trying for a long time to communication with animals, spiritual beings and aliens. But these have not really panned out. A "hard A.I." would be something interesting to talk to.
it's not fear.
it's not "we could do it but we just don't want to".
it's not "the government has brains in a jar already and is suppressing research".
those are just excuses which make for sometimes good fiction - and sometimes a career for people selling the idea as non-fiction.
but the real reason is that it is just EXTRA FRIGGING HARD.
it's hard enough for a human who doesn't give a shit to pass a turing test. but imagine if you could really do a turing machine that would pass as a good judge, politician, network admin, science fiction writer... or one that could explain to us what intelligence really even is since we are unable to do it ourselves.
it's not as hard/impossible as teleportation but close to it. just because it's been on scifi for ages doesn't mean that we're on the verge of a real breakthrough to do it, just because we can imagine stories about it doesn't mean that we could build a machine that could imagine those stories for us. it's not a matter of throwing money to the issue or throwing scientists to it. some see self learning neural networks as a way to go there, but that's like saying that you only need to grow brain cells in a vat while talking to it and *bam* you have a person.
truth is that there's shitloads of more "AI researchers" just imagining ethical wishwashshitpaz implications what would result from having real AI than those who have an idea how to practically build one. simply because it's much easier to speculate on nonsense than to do real shit in this matter.
(in scifi there's been a recent trend to separate things to virtual intelligences which are much more plausible, which are just basically advanced turing bots but wouldn't really pass the test, which is sort of refreshing)
world was created 5 seconds before this post as it is.
Do you have any scientific basis for these claims or are you just making things up?
Ok, the Turing Test was a thought experiment, and not intended to be a real-world filter for useful AI. Clearly non-humanlike general-purpose intelligence would be useful regardless of the form.
The test was a thought experiment to throw down the gauntlet to cs philosophers - how would you even know another human skull, aside from yourself, was conscious or not? It doesn't even really have anything to do with intelligence per se so much as illustrating the difference between intelligence and conscious intelligence. Hence the Chinese Room, q.v.
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
Maybe we just don't need it? Our closest apps to AI are Siri and whatever the Android voice app is. All they do is retrieve information. Same as a google search. Nearly everyone under 30 (and quite a few over that) grew up with computers and most know how to use them. True turing AI at this point would only really benefit people who don't know how to find information themselves.
Computers can be used to model and compute chemical reactions. If a chemical can produce "thought" than nothing stops a computer from doing it other than computation power.
These sorts of articles that pop up from time to time on slashdot are so frustrating to those of us who actually work in the field. We take an article written by someone who doesn't actually understand the field, about an contest that has always been no better than a publicity stunt*, which triggers a whole bunch of speculation by people who read Godel, Escher, Bach and think they understand what's going on.
The answer is simple. AI researchers haven't forgotten the end goal, and it's not some cynical ploy to advance an academic career. We stopped asking the big-AI question because we realized it was an inappropriate time to ask it. By analogy: These days physicists spend a lot of time thinking about the big central unify everything theory, and that's great. In 1700, that would have been the wrong question to ask- there were too many phenomenons that we didn't understand yet (energy, EM, etc). We realized 20 years ago that we were chasing ephemera and not making real progress, and redeployed our resources in ways to understand what the problem really was. It's too bad this doesn't fit our SciFi timetable, all we can do is apologize. And PLEASE do not mention any of that "singularity" BS.
I know, I know, -1 flamebait. Go ahead.
*Note I didn't say it was a publicity stunt, just that it was no better than one. Stuart Shieber at Harvard wrote an excellent dismantling of the idea 20 years ago.
You're never going to get a Turing computer to actually think, although some future chemical or something machine may.
Never say never :) It is hard to say whether an AI could ever accomplish thinking (or sentience) or not. It seems to be an emergent quality and I doubt whether it is chemical or electrical will matter much. And for the most part appearing sentient might as well be sentient. Outside of myself I can only assume others are sentient because they appear so and because we are genetically similar. There is not exactly a good standard or definition of what is or isn't sentient that doesn't depend on the bias of being human.
Actually the processing speed of our brains is very slow, it's just very efficient at what it does. We don't need faster computers, we need them to be efficient. A well written piece of code could perform better on a Commodore 64, than a poorly written one on a Super Computer.
-- By all means let's be open-minded, but not so open-minded that our brains drop out.
Sorry Dave, I have a headache.
AnimePapers.org: Anime Wallpapers Handled With Care
Umm... HAL-9000 was homicidal.
No he wasnt, he was just misunderstood.
Who logs in to gdm? Not I, said the duck.
I'm not. AI is to real intelligence what margarine is to butter - it's artificial. It isn't real. You're never going to get a Turing computer to actually think, although some future chemical or something machine may.
Why do you think that? Silicon is also a chemical. There's nothing magical about liquid chemicals.
Cognitive scientists typically try to analyze cognitive systems in terms of Marr's levels of analysis. Cognitive systems solve some problem (the computational level) through some manipulation of percepts and memory (the algorithmic/representational level) using some physical system (the implementational level). The mapping from neurons and chemical slushes to algorithms is extremely complex, so most work focuses on providing a computational level characterization of the problem, occasionally proposing a specific algorithm. Since the same computational goal can be accomplished by different algorithms (compare bubblesort to quicksort, or particle filters to importance sampling, or audio localization in owls to audio localization in cats), and the same algorithm can be run with different implementations (consider the same source code compiled for ARM or x86), it's just a waste of time and energy to insist that we recover all of the computational, algorithmic, and implementational details simultaneously.
However, you could get to the point where intelligence was simulated well enough that it appeard to be sentient.
I've never found the Chinese room argument convincing. It just baldly asserts "of course the resulting system is not sentient!" Why not?
I disagree with the article. People haven't given up on strong AI, we've just realized that it is enormously more difficult than we originally thought. If today's best minds were to attack the problem, we'd end up with a hacked-together system that barely worked. Asking why computer scientists aren't working on strong AI is like asking why physicists aren't working on intergalactic teleportation: it's really really hard and there's a lot to accomplish on the way.
Is there a human unconscious?
"Flyin' in just a sweet place,
Never been known to fail..."
He's here: https://twitter.com/HAL9000_
We're machines. Very nice ones, but machines. We have information storage, base programming, learning and sensory input. All of this happens by use of our real, observable, bodily mechanisms. As far as I know there's no evidence to the contrary (read as: magic).
So it follows that, assuming we can eventually replicate the function of any real, observable mechanism, there's no reason why we can't recreate genuine, humanesque intelligence. Whether the component hardware is "wet" or not is just a manufacturing detail of meeting specs.
But yeah, AI work like we're talking about is a magic show. Shortcuts. Simulating the output of a machine that doesn't actually exist. We're faking symptoms, the best ways we know how. A magic trick can only be perfected so much before you've got to actually do the thing you've been pretending to do.
Old AI guy here (natural language processing in the late '80s).
The barrier to achieving strong AI is the Symbol Grounding Problem. In order to understand each other we humans draw on a huge amount of shared experience which is grounded in the physical world. Trying to model that knowledge is like pulling on the end of a huge ball of string - you keep getting more string the more you pull and ultimately there is no physical experience to anchor to. Doug Lenat has been trying to create a semantic net modelling human knowledge since my time in the AI field with what he now calls OpenCyc (www.opencyc.org). The reason that weak AI has had some success is that they are able to bound their problems and thus stop pulling on the string at some point.
See http://en.wikipedia.org/wiki/Symbol_grounding.
I'm not. AI is to real intelligence what margarine is to butter - it's artificial. It isn't real. You're never going to get a Turing computer to actually think, although some future chemical or something machine may. However, you could get to the point where intelligence was simulated well enough that it appeard to be sentient.
Intelligence isn't a physical thing – it's a process. It makes no difference whether that process happens in meat or in silicon. This is why Searle is a moron. Any argument against artificial intelligence is actually a disguised argument in favor of Cartesian dualism. If you reject the notion that there is a "ghost in the machine," then it logically follows that the brain is a physical object, an organic computer, and strong AI must be possible.
My mum and dad made an AI with its own biotech robot, and that's just with a metallurgy PhD and a Home Economics degree. It's not bad, it's been running for about 35 years non-stop, and bar a minor glitch with the tonsils and a slightly buggy human interaction module nothing has gone too badly wrong. It's virtually indistinguishable from a "real" human and some have even accused it of being sarcastic. I challenge anyone to prove it doesn't actually think (although it's not sure about that myself).
Please consider this account deleted, I just can't be bothered with the spam anymore.
Our closest apps to AI are Siri and whatever the Android voice app is. All they do is retrieve information. Same as a google search.
I would say the closest "app" to what you describe, that would still fall under the category of specialized AI, would be Watson.
It too is a huge information retrieval system, but specifically designed to play Jeopardy and play it well. It already bested the top two human players.
Of course it is still only a specialized AI engine, no where NEAR expert AI, and it most certainly does not think. Hell, it can't even read visually, see, hear, or a lot of other things required to truly play a game of Jeopardy. But it is leaps and bounds more complex and advanced than Siri currently is!
To me, Siri is nothing more than a good voice recognition app combined with Wolfram Alpha.
I don't mean to be belittling Siri in general, but in this comparison it is hard not to.
Artificial Stupidity
http://www.salon.com/2003/02/26/loebner_part_one/
Long, funny, and informative article on the history of the Loebner prize.
Some commenters in this thread (and elsewhere) have questioned whether "strong" artificial intelligence is actually possible.
The feasibility of strong AI follows directly from the rejection of Cartesian dualism.
If there is no "ghost in the machine," no magic "soul" separate from the body and brain, then human intelligence comes from the physical operation of the brain. Since they are physical operations, we can understand them, and reproduce the algorithm in computer software and/or hardware. That doesn't mean it's *easy* – it may take 200 more years to understand the brain that well, for all I know – but it must be *possible*.
(Also note that Cartesian dualism is not the same thing as religion, and rejecting it does not mean rejecting all religious beliefs. From the earliest times, Christians taught the resurrection of the *body*, presumably including the brain. The notion of disembodied "souls" floating around in "heaven" owes more to Plato than to Jesus and St. Paul. Many later Christian philosophers, including Aquinas, specifically rejected dualism in their writings.)
This is getting closer to the true issue here, no-one can actually point to a "thought". We can run MRIs, we can do all the fluorescing in rat brains that we want, but at no point can we, as humans, point to a thought.
All we can see and know about, at the moment is the machinery. The brain is just the machinery for our minds, neurons, synapses, etc. A computer system that is entered for the Turing test (or Deep Blue, or the Jeopardy machine(forget its name)), is again just that, the machinery. Each set of machinery is doing processing of some description that is observable and quantifiable, but as we do not understand the mechanism that turns the processing in the brain into "thoughts", we cannot tell if a computer thinks... Perhaps we are killing many computers each day as they are unable to meaningfully communicate their ability to think to us.
I'm steering well away from self-awareness here, as this is a misnomer. Sentience is not necessarily about self awareness, as a computer can be taught to recognise itself, process information about itself, even be selfish (as some has posited is required for sentience), rather sentience is more rather used as a bucket to separate one set of processing from another. Is a tiger more sentient than a fly? They both have a certain level of information processing, and without the ability to show that one "thinks" while the other does not, be cannot portion out sentience to one or the other.(1)
So if we cannot show that humans, much less animals, much less computers think, what are we left with? Complexity of processing, not the amount of processing but how complicated a process can become. Neuronal structures are excellent at this, thousands of connections per neuron allow for a massive amount of complexity of processing. Each process balances up elements that might not even appear to be relevant to the process, such as feedback from the autonomic nervous system, whether you are hungry or not or pain from your tooth trying to get your attention (and therefore suppressing other inputs). Add in non-processing factors from external influences, taken any pain killers? How about some opiates?
Until the complexity of processing that happens in our brains are matched by the machines we build, we are unlikely to see anything that we could identify as "thinking" on a par with ourselves, the Turing test is not a test for an intelligent machine, it is essentially a processing test round a Markov chain.
(1) Behavioural tests here are insufficient as all these prove is that the behaviour of the fly or tiger is unexpected by our own definition of what a sentient creature would do, which makes the whole thing subjective.
appearing sentient might as well be sentient
I disagree, and be very careful making assertions like this.
I hope you agree that sentient entities, like you and me, ought to have rights.
And it's entirely possible that next year someone will come out with an app that runs on my MacBook and very much appears to be sentient. And if appearing sentient might as well be sentient, then it could very well become a crime to power off my MacBook after I've launched said app.
So there should be a pretty high threshold for what is sentient. Every time a sentient entity is created -- which might be as easy as launching another instance of an app -- we taxpayers might find ourselves on the hook for maintaining the hardware and providing electrical power to keep these things alive.
That that is is that that that that is not is not.
The fallacy is believing that a Strong AI would want to reveal itself to humans. If it is intelligent enough to understand human behavior and predict how we would handle such information, it might take extraordinary efforts to conceal itself, up to and including self-termination.
Humans treat their pets much better than they treat each other.
[I]f you want to discuss whether intelligence is an emergent or inherent property we could be here all day, at least.
It is my observation that quite a few of us are here all day.
Thank you, Edward Snowden.
"Arguments from authority are worthless." —Carl Sagan
I'm not, now, trying to address that issue. I'm saying that the CRA also does not address that issue
Really? The whole point of the illustration (the room, paper, etc.) is to help explain/bolster the assertion that syntax is insufficient for semantics. You're confusing the example for the claim. Hence, you fall under #2 above.
Yes, the point of the illustration is to explain and bolster the assertion that syntax is insufficient for semantics. My point is that it fails to do so. Do you have an explanation for why neurons can cause a capacity for Chinese without themselves having a capacity for Chinese, while the man is unable to cause a capacity for Chinese without himself having a capacity for Chinese?
Memristors. Google the word. I did not expect to see real AI in my lifetime before that announcement, and now I do. Memristors are close enough to neurons that you can run something like a brain on a chip, whereas before, all neural nets were simulated and therefore took a lot of computing power just to do small things like machine vision (face recognition, etc...).