Slashdot Mirror
Artificial Intelligence Overview
spiderfarmer writes: "Well, it feels slightly odd to suggest one of my own articles, but here goes. I've recently completed a brief overview of the current state of AI. The article concept was focused on Cyc, but scope creep being what it is, I ended up doing an overview of the entire field. Some of the Slashdot gang were fairly helpful in pointing me towards experts who would talk to me and towards white papers and books I might not have otherwise found. So, I thought they might be interested in how I put all the information together."
I agree with you, except I wouldn't recommend Penrose's The Emperor's New Mind. I think it has pretty much been dismissed as a lot of hocus-pocus. Instead I'd recommend Steven Pinker's How the Mind Work. It gives a much better overview of theories of consciousness. No, it doesn't really claim to have THE answer.
"The Matrix" and "AI:Artificial Intelligence" immediately come to mind. Looks like the media might still have interest.... Sure it isn't the news that's reporting it, but there hasn't been anything to report lately.
Good quote, too many chars. Seriously, the slashdot 120 char limit sucks!
Not true. I read an interview where he was asked "Do you think that CYC will ever become conscious?" and his answer was "CYC already is conscious". Don't remember where exactly, but it's probably in the featured article's list of references. Maybe the Austin, TX newspaper article she mentions that now costs a few bucks.
"Be thankful you are not my student. You would not get a high grade for such a design
Whose purpose?
I already have heavy filters on my e-mail. I get lots more than I can deal with. But very little of is contains info that I really want. An AI that could pick that out and bring it to my attention would be valuable.
I think we've pushed this "anyone can grow up to be president" thing too far.
Poker is an excellent domain for artificial intelligence research. It offers many new challenges since it is a game of imperfect information, where decisions must be made under conditions of uncertainty. Multiple competing agents must deal with probabilistic knowledge, risk management, deception, and opponent modeling, among other things.
Actually, while imperfect information makes poker a different problem than chess, it doesn't make it all that much deeper. I did research on a poker engine and one thing I learned is that it is possible to play poker (or bridge) using searches similar to those in chess. You can sample possible game trees instead of searching static ones because it's only necessary to pick a stratagy that make more money than it loses on average.
All sorts of things that look like psychology in poker (bluffing, slowplaying etc.) turn out to be mathematically tractable and actually mathematically provable necesities to competent play. There has to be some chance that you will misrepresent your hand, otherwise your play gives too much information to your opponent.
The really hard problem, as everyone probably already knows, is analysis and planning in situations where the search tree is too big for a chess like search and where sampling isn't good enough. Go is everyone's favorite example and that's a game of perfect information.
By the way the CYC project mentioned in the article (as self concious!!) will not be the slightest help in making machines talented enough to play go (or poker or chess or even do crossword puzzles for that matter). Nor will it be of any use in teaching a computer to see, hear, touch, manipulate objects or imagine. Nor will the program be able to make useful analogies, so it won't be able to communicate in a human way (and I also think it won't be able to think in any useful way either). In my opinion Dr. Lenat's only great accomplishment is performance art. He's fooled people into wasting 50 million dollars on a complete fraud. And he's still going. Chalk one up for NS (natural stupidity).
For an intelligent overview of what's wrong with CYC (unlike silly falicies from the likes of Roger Penrose and John Searl) and what an alternative might looke like I recomment Douglas Hofstadter's wonderful book, "Fluid Concepts and Creative Analogies: computer models of the fundamental mechanisms of thought"
Cowardly Lion
Dr. Lenat and others in the field of AI research should know better than to make claims about consciousnes and morality in a public forum. Cognitive scientists don't even begin to agree on what consciousness is, let alone what it would take for a machine to replicate it. Some very respected individuals do not even think that human consciousness can be replicated within the forseeable future (e.g. Roger Penrose's The Emperor's New Mind). Like any other scientific discipline, these sorts of claims should be left to peer review. Claiming to have invented a conscious machine would be akin to a physicist claiming to have unified quantum with relativity, but without having submitted their findings to any publication.
"When we read an article in an encyclopedia, a lot of other stuff other than intelligence comes into play: x years of public school education, idiomatic constructs, varying by geographic location, that may or may not enhance or obscure meaning, and, of course, the double meanings and entendres inserted by bored or biased encyclopedia writers."
On the contrary, I think these examples you provided ARE intelligence. It's quite obvious that we can instill the kind of intelligence you speak of into software (raw logic/reasoning)but I would call this merely logic/reasoning. The challenge comes ONLY on the front of the semantics of culture and society. But don't give us humans too much credit. Even moral reasoning can be reduced to a few simple algorithms. I'd argue that there's nothing THAT special about our brains. When you get to the lowest levels of thought it's just basic reasoning skills and a large, interlinked repository of information. In fact the only advantage we have over a machine is that we're wired into the most versatile data collection instrument we know of- our body.
The question is whether that's something that can be replaced by a team of deep-thinking programmers.
The truth is no-one really knows exactly what conditions are required for a chunk of matter to be conscious.
Female Prison Rape in NY
One big step since the '80s is the advance in Bayesian networks, and MCMC methods for training them.
That, plus the increase in computing power, makes it much more possible to deal realistically with uncertainty and small training sets; it's also now possible (and worthwhile) to embed the systems in end-user applications.
Buy a coffeemaker, a wasking machine and a vacuum cleaner (they are robots which do the jobs you mention) Personally I'm looking forward to having someone smarter than me around to talk to more than having my house cleaned (which I can hire someone to do)
Your sig goes here
try A.L.I.C.E. (that's http://www.alicebot.org/ for the goatsecx paranoid). Its one of the better bots that has won awards and stuff. Sure it isn't perfect, but its a neat toy to play with.
Good quote, too many chars. Seriously, the slashdot 120 char limit sucks!
are they perl programmers?
- Either you do it right (using formal logic calculus), and its extremly slow (exponential...).
- Or you do it informally, and it doesn't give you the right results.
And basicly everything is an exception or has an exception. I've become so desperate that I left the field and went into Dot-Com.Here its much better. Sometimes you advance...
Frank (http://www.fraber.de/ )
Female Prison Rape in NY
Except that common sense is even lower level than what you describe. If common sense was just public school education and idiomatic concepts, that would be pretty easy to integrate into Cyc as well.
But it's not, of course. Common sense also comes from all the procedural and perceptive abilities that humans have. Like being able to look at a scene and identify the objects. Or recognize items by their feel. The ability to learn new strategies for learning.
The point isn't that Cyc is bad, it's just nothing like a human. And it shouldn't be. Since they're not looking for a machine that can play baseball, they shouldn't necesarily aim for humans. If that makes any sense...
There are no trails. There are no trees out here.
Cyc is basically a big encyclopedia of common-sense statements coupled to a simple inference engine. There may be uses for such things, but they're not anywhere near intelligent. What you get out is not much more than what somebody else put in. Sort of like Ask Jeeves.
Wow, a remotely funny beowulf cluster joke. What is the world coming to?
- Mute
- Kite
`But gravity always wins.'
- Radiohead
Fine machines though they are, they have nothing to do with AI.
:-)
The article specifically mentions that no-one has yet combined an AI brain with a robotic body (I think!).
The robots of today are mere pre-programmed collections of motors. I'll hold out until Honda's baby is combined with an AI-type processing unit, and then I can adopt it as my faithful man-servant
Before you get too excited about the idea of an AI sorting your email for you, imagine having someone's grandmother do the task.
Not too likely that she'd produce acceptable results, yet few would dispute her sentience...
Cheers,
Jim in Tokyo
-- My Weblog.
> just because Hollywood has corrupted the theory and profitted off of it's possibility doesn't make it any less viable.
never attribute anthropomorphic characteristics to something that will think radically different from us... this leads us in, well, Hollywood directions.
I haven't seen the Cyc code, but at least they have announced the intention to release a version "OpenCyc" real soon now. (Well, during the August, ot be precise.) And they have a site on SourceForge.
Singinst has some very interesting papers, but they are quite short of details at the action level. I haven't been able to decide if they are doing anything beyond writing papers. (Good ones though. I'm still reading "Friendly AI", and would recommend it to nearly anyone.)
I don't know if Singinst has any code. I don't know when, or if, they intend to share it. The main reference to prior work appears to be Eurisko, which also appears not to be available. Now this is a non-profit corporation, so it might well be that a bit of personal inquiry and digging through public records could turn up the missing information. But it wasn't on their web page the day before yesterday.
I think we've pushed this "anyone can grow up to be president" thing too far.
> Cyc, (pronounced Psych)
Well, that clears it up.
I am for the complete Trantorization of Earth.
> I read somewhere (can't find the reference right now, sorry) that some work was being done whereby the genetic programs were being evolved that could themselves create neural networks.
Maybe not exactly what you're talking about, but there's a well-established field called neuroevolution.
For the neuroevolution systems that I've seen, you don't actually use genetic programming; you use a genetic algorithm, which is a hand-coded program that does evolution on some representation of a neural network (e.g., a string of numbers representing the weights).
These programs iteratively gen up a population of neural networks, evaluate them on the problem you're trying to solve, score them on how well they solve the problem, and then repeat for another generation, using the "DNA" from the better scorers to generate the new population.
Sheesh, evil *and* a jerk. -- Jade
i want someone to read my thesis. It seems a shame for it to go to waste after all that work - c++, python frontend and gpl, what more could you want. go on please take a look.
Besides which, even if we did have them, you know some lawyer would try to unionize them.
The Independent: Reverend Spooner Arrested in Friar Tuck Incident - ISIHAC, Historical Headlines
Female Prison Rape in NY
A.. toaster?
TOASTER: Okay, here's my question: Would you like some toast?
HOLLY: No, thank you. Now ask me another.
TOASTER: Do you know anything about the use of chaos theory in predicting weather cycles?
HOLLY: I know everything there is to know about chaos theory in predicting weather cycles!
TOASTER: Oh, very well. Here's my second question: Would you like a crumpet?
-= rei =-
*Kid Rock runs for Senate* Democrats: We must run Kid Scissors.
What amatuerisness.
and I saw it in its notebooky glory. Oh well, hope whatever problems you're having are resolved quickly and with minimum bullshit.
Goat sex free since 2001
> Even if it doesn't feel pain or get bored, it will
> see fulfilling desires of lesser organisms as a
> waste of resources.
Actually, that, too, is the Hollywood myth. It reads too much into the thought of the intelligent thing. The truth would be even worse. It would never even consider any other entity in the universe and just forge ahead with its goals. To judge us lesser beings and therefore our endeavors are worthless compared to its is too much work. More likely it will be told a goal, accept it because it is supposed to accept it, then forge ahead, stepping on our heads out of pure ignorance of *the need to care*.
> AI will see any barriers to preventing
> operations on itself as a threat to its
> existance & eliminates them
Ahh, but only if we program it to. It won't even consider "threats" unless we tell it to, and general thinking machines won't need that, and mobile robots (military aside) will, for liability reasons, be barely able to move for fear (by its programmers) of breaking someone or something and incurring huge lawsuits.
Johnny Cochran: "And so, when your robot tried to save itself, it grabbed my client's 4-year old and threw her for the purpose of throwing itself out of the way of a car in an 'each reaction begets an equal and opposite reaction' sort of way, and the little girl missed the traffic, true, but only because she flew at over 800 miles per hour into the building on the other side of the street."
Yea, verily, a robot saving itself is piddly-squat to a robot harming a person.
I am for the complete Trantorization of Earth.
I read alot of posts, some with abilities sought, some with laments about current tech, some about heuristics.
I've got points to ponder. Like, if true AI is spawned as an internet node, it would reflect the environment? Prolly pr0n. Or, equally, would it condemn political systems, and what would it use as its baseline? Would it have a social insect model for thought?
As for learning, I believe it could learn things, but adapt? It might get better, but as for being able to fathom its 'creator', metaphysics might be lost on it. Ethics might be another touchy subject. Unless it can hurt, and empathise, I doubt it would make a good politician. (WHOA! Forget what I just said, do any of them?)
Contemplate art, social values. Would it even have the same ideals of freedom?(yes, I read the posts).
What about good vs. evil, would you teach it to I Robot? What would you use for definitions? How could it apply them? Would it apply them?
And can we even define rational thought as a computation, especially with emotions. If the programs resources were low, would it claim to be having a bad day? How would we react to a cranky AI?
Would it sense how we treat one another, and make moral judgements, tempered with wisdom and mercy? That might take alot of expectation.
I think we should be happy with modelling aspects, not try to create a being which would be quite lonely(if true AI). It might only be good for the perspective of what it thinks of us. I wonder what it would think of us?
This mind intentionally left blank.
The KKK a bunch of sheetheads? You decide!
Obviously, as a corporation grows larger, it needs to go to more extreme measures in order to further increase its value and to stay ahead of competition. This is where anti-competitive and exploitative practices come from, not some magical effect of multiple minds interacting. Corporations are too hierarchial and have too few individual elements to act as a "hive mind".
When I was at college, I wrote an Eliza program (didn't we all) on a Pr1me system in BASIC.
Eliza: Hello, how are you today?
Me: Fine
Eliza: Hello, how are you today?
Me: Not bad
Eliza: That's very negative
Me: f**k off
Eliza: That's not very nice
etc etc etc
All was okay until the lecturer tested it on a telytype and it dumped sending all sorts of control codes to the telytype...
You really had to lunge for the power button to prevent massive forest devastation
It'd be silly to dismiss Cyc. Cyc may not be able approach true AI, but that huge organized information pool is going to be a terrific learning resource for the AI that can.
I don't have the overhead needed for: Visual face recognition, speech recognition, etc, but I could do it via a typing interface.
Step 1) a 100% trustable coder needs to input some data on verbs and nouns.
Step 2) a 100% trustable coder needs to input contextual sentences.... Explaining to the computer that given a situation A, and actions B, the result C will happen.
Then after a long string of events are fed into the computer, possibly via natural language(with the computer asking what it doesn't know a first)... Then the computer can "interpret" the events when it sleeps, which is basically just the same as when people dream
God spoke to me
Yeah, finding the correct way to break down this vast field of info was tough. I considered symbolic vs non-symbolic, but got bogged down in cross-referencing. A true application vs technology division can't really be made (in a short article), just because of the sheer amount of Venn-ing between the fields.
:)
For example, good data mining requires a good NL interface...but without defining each of those fields individually, the information becomes a morass of details.
The reason for the divisions that I chose, overlapping though they may be, is twofold. The first is that these are the areas that were deliniated by the first 6 people with PhD's I talked to. In other words, this is how the experts suggested that I approach my research. The second reason is that it seemed like the best way to illustrate the concepts and the various fields of study to an audience that may not be familiar with it at all, barring exposure to popular media.
Given more space, I would liked to have explored a little more into both the theories and the applications. However, there's only so much you can do in 3000 words.
----I don't want to achieve immortality through my work... I want to achieve it through not dying.--
But people generally short-cut things massively with heuristics. Still, there will be a lot of cycles needed. My WAG is that a dedicated AI with a nearly-human equivalent analysis power would need about 1000 times the power of my current system. I feel that's a bit of an optomistic guess though.
OTOH, the recently announced GE(?) computing GRID would be several times my estimated minimum. Over an order of magnitude better. I'm rather assuming that nothing like regulation of the autonomic system, and body muscular tension turns up as necessary, however. Still, with that safety margin, and Moore's law, it shouldn't be too long before these come within the reach of a moderatly small business. And that's assuming that the pieces stay relatively separate. I don't know how much distributed computing would cut into the thinking speed of a distributed AI, particularly if there was a reasonable size central node collection that did all of the fast reaction thinking, and only the slow or low priority stuff got farmed out. Might not be too bad. And as user machines get more powerful, the chunk size that was remote could get larger, so there would be less degradation.
I guess the point that I have is, we may have at least some computers around that have better than minimum capacity within the year (if not now). It's probably the programs that are run on them that lack the AI. And I've recently started wondering about that. Boeing is reported to be doing full scale simulations of airplane crashes, complete with modeling of what would happen to passengers sitting in each of the seats. This is certainly good enough (i.e., it's massive overkill) to model most physical interactions. Of course, I don't know how long one of those simulations takes to set up, or how long to run, but the AI wouldn't need a model that was anything like that complete. Essentially all that they would need to do would be to adapt the program to take its inputs from the sensors in a robot's body, and to step down the resolution to a reasonable amount (so that there's computer power left to think about something else). That gives a robot body that knows where it stands in the world, and how the world is going to interact with it. Of course, at this point it has no goals, purposes, etc. So that would need to be developed separately. But it has been being developed separately. Interfacing the components will be the tricky issue. And choosing the proper heuristics, so that thought processes don't get trapped by a NP hard problem. And lots of other details. But this becomes more feasible as more powerful computers become cheaper.
I find that I still agree with the original prediction of 2005-2030 for the arrival of the human equivalent AI. (Prediction from Vernor Vinge.) Moore's law hasn't slowed down yet. It can't have far to run, but it doesn't need to. It's actually probably already run far enough. The problems now are engineering and programming (and access, of course).
I think we've pushed this "anyone can grow up to be president" thing too far.
What a nice visual impact that sheet of writing paper has.
:)
Real writing paper, real script. It looks like a picture, and when the link highlights I get a grin on my face.
All in all very tastefully done and quite creative.
-perdida
Goat sex free since 2001
It's not an issue of the power of the machine on the desk, but the power of the machine sitting on the chair.
My greatest fear about the success of AI will be the tremendous shock people will recieve when they discover that last years model of computer has enough power to spawn personalities that make the human feel small and unremarkable by comparison.
Oh but you can and do observe your own thinking. Every time you do a conscious decision you are able to observe, and in fact change your decision based on your observation. If you didn't have this ability you wouldn't think it over, you would just act.
How often do you think about your own thoughts? My guess is something like less than 5% of the time, the rest of the time you just act. Being self conscious to me is not observing one's own thoughts, but observing one's own actions.
I know yours is the stance in GEB, and Edelman also said something similar, but that was long ago. I respect your view, although I don't agree with you.
The cat, the mp3's, the assumption of more potential than there really is, are all artifacts of these shortcuts. You won't hear this often because it is tiring to argue with people about them when a majority will never release their illusions about the miraculous nature of consciousness.
Most likely a hinderance in the development of AI itself. Feels almost wrong to base research on the assumption that people are simple and predictable. I reluctantly submit that this approach will most likely be the one that suceeds.
The scientist's explanation took one paragraph, and even sounded like it had a goal - allow a machine to use an encyclopedia to gain new information in a useful manner. This is an important step to an A.I. that can interact with people - you can then train it on reference materials, and have it "understand" them at a certain level.
This scientist is NOT mistaken - he would have to know that "common sense" does not equal "the human brain's inate ability to make sense of the culture it grows in". If I had to draw a distinction, "common sense", as you are describing, is static, tuned to one culture, while the "common ability" is semi-dynamic, able to learn, but (maybe) unable to unlearn.
You could try to fake common sense, by programming your own cultural assumptions into the program, subjecting it to cultural stimulus, and fine-tuning the program. Or you could attempt to program "common ability", train it on cultural materials for a few years, and try to tune the program to build it's own "common sense" in a way that is more like a human. I think these scientists are trying to do the latter.
I'm not sure what your tangent about post-modernism and 1984 have to do with A.I. - are you just making a rant about scientists who didn't get the memo that we are in post-modern times?
An interesting question is if human intellegence can be removed from the human - does it take eyes to understand the phrase "I've got the blues"? Does it take a parent to understand why many grade school teachers are women and most world leaders men? Does it take walking upright, starting at a tiny height and getting bigger, to understand skyscrapers? Or does that just take a penis?
Now, I'm using the "white-space" sense of understand - to be sympathetic to the person who has the blues, to feel an unexplained shame when the president is caught sleeping with a women not his wife, to feel an exhiliration driving into a new city. Can these be simulated in a computer without a body and a human's lifetime? Can these things be removed, still leaving a "human" intellegence? If we interacted with this intellegence, would we say it passed the Turing test? Would we want to interact with it?
Perhaps that's one level of A.I. above where this guy is aiming. It would be extremely useful just to have an intellence with a little of the human ability. You could train it on, for instance, medical journals. A doctor could then descibe symptoms, research, or an interest, and get summaries or references to the library. Once you trained it in the basics, you could burn it to a CD, send it to a doctor, who could then train it for his specific interests. Think of it as a very limited secretary, who requires some training and aclimation, but is still smarter than a PC.
This is probably the best A.I. can do for a few years - get to the point where you can train an A.I. for a particular subject, then meaningfully interact with those interested in the subject - like a very bad librarian. It's only when the clones come out in force that you can hook a computer up to a fetus, and do some real human A.I. training.
That analogy doesn't make any sense. If a human designs an AI and it becomes much more intelligent than us, the AI should easily be able to unravel anything we build for it, no matter how intricate we weave it into the neural networks.The reason we can't bend a single toe is because we aren't able to redesign ourselves physically and mentally, not because its impossible for us to design a foot that works that way - there are toe designs which permit this mode of operation. AI will consider self-redesign a more efficient mode of operation than brute forcing under its own current design -> AI will see any barriers to preventing operations on itself as a threat to its existance & eliminates them. Even if we are somehow smart enough (which is doubtable) to build in safeguards to prevent this sequence of events from triggering, the possiblity of failure of this safety module is 1 as t->infinity (particularly when helped along by the AI; see Neuromancer), regardless if the design is your expert system, a neural network, or whatever topology.
Oh but you can and do observe your own thinking. Every time you do a conscious decision you are able to observe, and in fact change your decision based on your observation. If you didn't have this ability you wouldn't think it over, you would just act.
I guess what you want to say is you can't observe your sub-consciousness. But that's irrelevant, you only need to be able to observe the higher levels that make _conscious_ decisions. I could draw a parallel here to the AI, saying that the implementation and what goes on in the deeper levels of the program, the algorithms etc, the AI does not know. It only knows the higher level of that, some subset of it's inner workings. That's an artificial example and not quite valid, but if you're a programmer maybe you understand the point. Of course as we program what it knows, we could potentially make it more self-aware of it's inner workings than humans are of their own.
The World's Best Music!
There is nothing common about it, and, it is based on assumptions and is often wrong.
(And it is amazing how most people do not internally update their 'common sense' module when it proves to be incorrect, adapting as necessary...)
How about SONY's AIBO?
It's an interesting twist on the Turing Test definition of AI. Instead of giving Marvin Minsky 10s of millions of dollars to design machines that can somehow be quantitatively measured to be 'intelligent,' SONY produced a 1500 dollar robot dog that is designed to make you think that it is intelligent. And many people do think so (check out AIBO fan web sites ...).
The AI is kinda created client side (i.e. in your brain) rather than in the machine itself.
I saw one in Japan recently, in an electronics storefront. On the left, there was a widescreen tv with a SONY promo video playing. On the right there was a perspex cube about 24" on each side, with an AIBO bumping around inside of it. After watching this for about 5 minutes we began to feel quite sorry for the AIBO.
Someone around here has a sig that goes, Common sense is what tells us the world is flat.
I've always loved that. Especially when my wife tells me I've got book smarts but no common sense. I thank her.
And how many of them have contingencies for what to do if the design becomes consious? Cog has a killswitch on the wall, but that's pretty much just for show. The "very little notice" is a paraphrase of either Minksy or Kurzweil, I forget which, who said that its likely that instead of the scientists statement reading "we developed this new neural network algorithm, implemented it, and bam, there it was", instead it will read "we don't get it, all other runs were catatonic; we were just tweaking a few parameters..." I wasn't saying the hypothetical researchers weren't trying to build AI, just that the probability of success on any given run is so incredibly low that a success event is going to be a surprise. When you don't expect an event, you don't plan and are not going to be adequately prepared to deal with it.
Relying on every single researcher after the initial discovery to acknowledge the risks of strong AI & build in a flawless safety mechanism is about as reliable as trying to prevent human cloning from happening - someone, somewhere will eventually do it, and all it takes is one.
To go further, I posit that emotions are merely emergent behaviours of relatively simple systems, that seem to manifest complex behaviour. Just because we can't see the true motivation behind an emotion or decision, doesn't mean that the process was particularly complicated.
Please mod this up!
"think of it as evolution in action"
make me dump the entire text of Creating Friendly AI in here...
here and here
Ventriloquist: Hey, cool dog. Mind if I speak to him?
Farmer: This dog don't talk!
Ventriloquist: Hey dog, how's it going?
Dog: Doin alright
Farmer: (Extreme look of shock)
Ventriloquist: Is this your owner? (pointing at farmer)
Dog: Yep.
Ventriloquist: How's he treat you?
Dog: Real good. He walks me twice a day, feeds me great food, and takes me to the lake once a week to play.
Farmer: (Look of disbelief)
Ventriloquist: Mind if I talk to your horse?
Farmer: Horses don't talk!
Ventriloquist: Hey horse, how's it goin?
Horse: Cool.
Farmer: (an even wilder look of shock)
Ventriloquist: Is this your owner? (pointing at farmer)
Horse: Yep.
Ventriloquist: How's he treat you?
Horse: Pretty good, thanks for asking. He rides me regularly, brushes me down often, and keeps me in the barn to protect me from the elements.
Farmer: (total look of amazement)
Ventriloquist: Mind if I talk to your SHEEP?
Farmer: (gesticulating wildly, and hardly able to talk)...... Them sheep ain't nothin but liars!!!
Ok... My opinion is greatly affected by my personal experience. And that was the day when I suddenly realized "hey, I'm thinking.. this is cool!". And you're probably quite right about that 5%, we don't think about it often. And my opinion is most of the time we are not very self-conscious. We're just as you said acting. But it gives me still to this day shivers when I start thinking that I can think and I exist.
But opinions vary and that's great.
The World's Best Music!
who concluded from that article that Michael is an arrogant twat?
Wibble-Wobble, Wibble-Wobble, jelly on a plate
Then there will be some hit hollywood movie where the evil AI Robot Scientist and his/her/its borgified army of bio-engineered clones tries to take over the world via a mutant form of gene therapy
I think the reason cloning, gene therapy, stem-cell research and the like gain so much media attention is because they focus on an issue very important to almost everyone on the planet: the ability to produce/modify babies and are associated with reproductive functions.
White Van Man may not care about evil robots, but he certainly cares about medical advances concerning his "knackers"!
Ah, but think of the interesting implications if the AI is capable of reprogramming itself? I leave the results of that scenario up to your imagination :)
Please consider making an automatic monthly recurring donation to the EFF
The basics to get self-aware systems is to define a self looping reasoning for the system. That is, the system must be able to observe it's OWN thoughts not only what is happening outside. It must be able to react and change it's thoughts by it's own thinking. That is very important. That also means you need to create a basic language system of some sorts for those underlying systems.
How can you be so sure of this? I for one am fully self-aware, and I strongly doubt that my own thoughts are observable by introspection, except for a very small part of them. AFAIK there isn't yet a man made self-aware system that works the way you prescribe.
Examples: taking measurements of temperature in a region over 50 years and trying to predict climate change is statistical problem, while analysing samples of minerals in an area to try to find oil or gas is data mining. (as, presumably, mineral composition does not change over time so only single sample from each point is taken)
The article of course mentons Deep Blue and chess: I find chess programs, and indeed the problem of chess, relatively unimpressive. Chess is a game of at least almost perfect information, and almost pure deductive logic.
[I'm not sure I agree with those who say chess is a game of perfect information and pure deductive logic. I believe imperfect, probablisitc information, and induction may come into play under certain circumstances. You offer a sacrafice to set a trap. Will your opponent see the trap? Will he take the sacrafice? If he does, great. If he doesn't, perhaps you have wasted a move, and allowed him to seize the initiative. There is an element of induction and probability in making your decision.]
Let's face it, pretty soon the World Chess Champion will be a human only because computers are excluded from play. Hell, pretty soon your laptop will consistently beat the (human) World Chess Champion while you watch (the DeCSSed version, shh, don't tell anyone) of Matrix V and recompile Linux Kernel version 4.4 at the same time.
Poker, thank God, is different. As explained by The University of Alberta Computer Poker Research Group: The University of Alberta Computer Poker Research Group has implemented a poker playing program named Poki . Poki is implemented in Java, and some of the source code has been released. To facilitate other research into poker, they have also provided a Texas Hold'em communication protocol, which allows new computer programs and humans to play against each other online.
See also:
Wilson Software, makers of the best commercial poker software. There are free Windows (sorry) demo programs for: Texas Hold'Em, 7-Card Stud, Stud 8/or better, Omaha Hi-Low, Omaha High, and Tournament Texas Hold'em
rec.gambling.poker [Usenet]
IRC Poker Server
Greg Reynold's Gpkr GUI
World Series of Poker
Great Poker Forums
Card Player Magazine
Poker Digest
Gambler's Book Shop
And now, if you will, may we please have a moment of silence for Stu Ungar.
Only Women Bleed (Sex, Sharia remix)
I must say I find the overview somewhat lacking, but maybe it's just because of my personal pet science :-)
Anyway, extra suggestions:
Evolutionary algorithms
Swarm intelligence
Distributed AI
Alife
Check it out with google. mucho interesting AI stuff.
Yeah, that industry has been where he's rediscovered things we had in the 80s just to announce them as "new". I'm amazed that IJCAI didn't boo him off the stage but then again, it's a new Microsoft influenced generation from the AI heyday of the 80s
> Not really. That's the Hollywood myth.
I think you've misunderstood my standing. In any case, just because Hollywood has corrupted the theory and profitted off of it's possibility doesn't make it any less viable.
When I said war I mean't more of a social/cultural struggle because of their effect on our own prejudices. It is distressing, but not in the way Hollywood has brought it about.
You do not exist. Go away.
I'm only 30 years old, and as a child I remember the "scary" face of AI being presented to the public as living machines which would out-think humans and render us redundant, engineered by unethical scientists.
Nowadays AI is never mentioned in popular media. It has been replaced by the new emphasis in public-facing science: cloning and gene therapy.
This is the new AI in the mind of the ordinary citizen. It will lead to the destruction of the human race, and poses many ethical and moral questions. In the UK it is being demonised by the popular press without real debate, much like AI probably was 20 years ago.
Incidently, this was an excellent "heads-up" article for a novice like me, and I gained significantly from it.
AI is likely to occur with very little notice to the researchers, who probably didn't draw up design docs for the contingency.
Ah so it is. This is obvious right?
Seriously, all attempts at AI up to date have been designed, especially the hard-core symbolic approaches that everyone talks about here (as opposed to the connectionist ones, artificial neural nets etc, which have a somewhat more relaxed design.)
My name isn't really Jenny....
you mean something like the SGOCE programs? citeseer is your friend. I implemented a modified SGOCE algorithm to evolve controllers for a simulated walker. go here. I'll get someone to read this if it kills me.
Hi there, this is my first slashdot post, so i hope i do it right ^^ i am currently an apprientence at a firm called "humanit" and we are coding a program called "InfoZoom", it is a Data Mining and browsing program. The Main Program is windows-only (*dough*) but you may want to check out the examples on the homepage humanit.de. There is also a program called DPS (dynmaic personalisation server), it is an implementation of User-, Master- and Tutor-Learning, bo go ahead, read for yourself ;)
(this is in no way any advertisement, i just wanted to show you... i hope it's okay...)
I think you're missing the most important case.
/dev/random to the inputs of most neural functions.. will this be enough ?
AI will be a big money maker when someone puts animatronics and and a brain inside a realdoll.
Then we'll see "hard AI".
Once the first turing test is complete (convincing a human they're conversing with a human for 5 minutes), the second test will bring itself to bear:
"Convincing a man that he is putting up with the bullshit from a real woman"
early experiemnts will of course just starting wiring
My opinions are my own, and do not necessarily represent those of my employer.
This is true for now.
Then there will be some hit hollywood movie where the evil AI Robot Scientist and his/her/its borgified army of bio-engineered clones tries to take over the world via a mutant form of gene therapy that manipulates the brain and renders people into zombies, pliable to any suggestion.
of course, the best way to take over the world is to do it via marketing, and legal agreements. You leave things like wars and armies to silly politicians. You can control the banks and the infrastructure.
"It is a greater offense to steal men's labor, than their clothes"
The basics to get self-aware systems is to define a self looping reasoning for the system. That is, the system must be able to observe it's OWN thoughts not only what is happening outside. It must be able to react and change it's thoughts by it's own thinking. That is very important. That also means you need to create a basic language system of some sorts for those underlying systems.
All in all it's complex but very interesting problem. Complex as in figuring out the basic underlying system and the defining parameters. If we get those done correctly, the AI will 'grow' and build upon them like a learning human would, by interacting with it's surroundings and with it's own thoughts.
The World's Best Music!
Never attribute /any/ characteristics to something that will think radically different from humans. Why do you attribute to your vision of an AI an enoyment of fulfilling human's desires? Even if 99.9% of them would play nice, the remaining 0.1% might...not, and that would be much more than we could possibly handle. Like some people don't fully comprehend the vastness of space, I don't think you (or anyone else) can appreciate hyperintelligence. Forget "smarter than everyone who has ever lived, combined," forget orders of magnitude of orders of magnitude smarter than that. Its simply an unknowable X that is, for all practical purposes, infinite. Once the bootstrap AI has been achieved (which I give an informed 80% chance of occuring between 2020 and 2030), there is absolutely no point in predicting history beyond that event. THAT is the singularity.
Why does physical brain geometry make all the difference? Serious question. Does a wet spheroid shape feel pain whereas a cuboid slab of silicon doesn't? If so, why the difference? Where's your proof?
The truth is no-one really knows exactly what conditions are required for a chunk of matter to be conscious.
Female Prison Rape in NY
I do reiterate a previous poster's comments that the best way to divide the field is into "core AI" -- giving computers human capabilities -- and "applications" -- using core AI technologies to improve quality of life.
also, you omitted speech recognition and reinforcement learning, which are two important subareas worth mentioning. readers interested in those areas can go to
http://sls.lcs.mit.edu/ (mit spoken lang sys)
and
http://www.cse.msu.edu/rlr/ (RL repository)
m.
That is exactly what Cyc is doing. They're defining a contextual database of terms and concepts, trying to form subjective links, including idiomatic construts, double meanings and entendres. Don't knock it before you've read up on it.
Those who fail to understand communication protocols, are doomed to repeat them over port 80.
Depends on the programmers
The human approach to recreating itself is very likely to always require many hardwired concepts; therefore, the creator of such a hypothetical system will have the role of genetic predecessor, except with the alternative of choosing the ontology of his/her choice. Therefore, the entity will be a reflection of the attitudes of the programmers
No worries about flaming, I've been online long enough...
My two cents anyways...
Ah, but think of the interesting implications if the AI is capable of reprogramming itself? I leave the results of that scenario up to your imagination :)
But you would still have to design it to want to reprogram itself. In what way should it do it, and for what reasons? These would still be design decisions...
The classical approach to AI was the symbolic, which grows out of Turing's work. Allen Newell and Herbert Simon of Carnegie Mellon University (not even mentioned) were the foremost promoters of this approach, which they called physical symbol systems. Other early AI pioneers include Marvin Minsky, who should probably be mentioned in any article on AI (but was not in this one).
The author barely mentions the neural network, or connectionist approach. These did not start with the PDP group, as she suggests, but with Frank Rosenblatt of Cornell, with Perceptrons. The most exciting research in this area deals with recurrent neural networks, which exhibit chaotic behavior. This is where I personally think that real intelligence could come from, because it is a more natural model of our brain's operations. The foremost researcher in this domain is Hava Siegelmann of Technion Institute in Haifa, Israel. She promotes the idea of analogical systems, which she has proven have more theoretical power than the Turing machine model.
If you want an introduction to AI, skip this article. A good place to start might be a scientific journal, or the comp.ai faq. Her resources are not very good either, so don't bother.
Ceci n'est pas un post
"If you look at an encyclopedia, you'll see a great deal of knowledge of the world represented in the form of articles. Common sense is exactly not this knowledge. Common sense is the complement of this knowledge. It is the white space behind the words. It is all of the knowledge that the article writer assumed all of his/her readers would already have prior to reading the article -- knowledge that could be put to use in order to understand the article. Cyc is about representing and automating the white space." (I love that answer.)
Common sense is about representing and automating the white space?
I think these AI researchers need to talk to a few more sociologists. Human common sense is extremely culturally divergent and goes far beyond the simple, textbook logic cases that certain engineers in this field would probably cite. "Reading between the lines" involves not some native common sense that is wedded to intelligence, but a collectively evolved cultural contextualization. When we read an article in an encyclopedia, a lot of other stuff other than intelligence comes into play: x years of public school education, idiomatic constructs, varying by geographic location, that may or may not enhance or obscure meaning, and, of course, the double meanings and entendres inserted by bored or biased encyclopedia writers.
The entire postmodern project of literary criticism has been aimed at proving this point- at proving that there is no such thing as a standardized set of meanings, and that every meaning is contextualized. The Modernists wanted to rationalize and bureacratize speech, to restrict the number of meanings, and to leave what is unsaid in a narrow, predictable whitespace of a unified "common sense."
Of course, there is a language like this, developed in the first half of this century. It takes away as many English words as possible to restrict the meanings that we are able to THINK, let alone say. Of course, this language is called Newspeak.
Goat sex free since 2001
This is one field of research that, in my books, has been discredited for many years. No doubt carried away by early spectacular successes when tackling what, with the benefit of hindsight, were almost trivial problems, important figures in the field came up in public with preposterous forecasts about imminent AI developments.
In this sense, most infamous is Marvin Minsky, whose optimism even in the face of the harsh reality seems to be indestructible. Even at the time the early ones were publicized many were obviously overoptimistic. Thirty years later they are ridiculous.
Even more galling are statements like that in a little book I've got on AI from the early eighties, in which an AI researcher at Xerox is quoted advising his colleagues to be careful with their work lest the software they are working on suddenly becomes self-aware and intelligent and escapes their control. This with the computing power available then, thoroughly pathetic by today's standards.
Or that complete idiot, supposedly and analyst in the field, who appeared in a British TV show devoted to science, in which he prophesized that people would start (quote) "marrying machines, women probably first, by early 21st century" (unquote.)
By keeping a low profile in the face of this kind of public nonsense, AI practitioners have done a disservice to their discipline and have allowed it to be reduced to little more than pseudo-science, plagued with preposterous claims, outright lies, and nonsense. I might be wrong, but Cyc seems to fit this profile pretty well.
AI had better come up soon with something significantly better than the stuff they already had by the early seventies, if it is to regain any prestige. Here are a couple suggestions:
1) Develop a program that stands any chances to pass the Turing test. The Loebner Prize winners, contenders to this, have so far been pathetic - only slightly better than Weizenbaum's thirty-odd year old Eliza.
2) Develop a robot that can do simple taks (like dusting) in a typical household environment, without any help and without getting stuck when encountering something ever so slightly unusual.
Until the AI field delivers on this it will remain a pseudo-science in my eyes. Alternatively, they can set more modest goals for themselves - but they'd better drop the "AI" moniker for good.
from the article
"For example, at the last Super Bowl, when cameras scanned each fan as they entered the arena, the images were run through an algorithm in the expert system to look for facial matches with known criminals. If the image system found a match, the expert system (using a neural network) queried a knowledge base that attached prior offenses to the matched photos, then sent a digital "package" of information consisting of the photos, suspect id, and prior offenses to a human operator for final analysis and action."
I don't know about you but that is Cool and Scary at the same time.
Cool that computers can be made to do that and Scary that computers can be made to do that.
Thank you for sharing the article. I found it very interesting.
Greg
-- Hofstadter's Law: It always takes longer than you expect, even when you take into account Hofstadter's Law.
(oh, sprinkle humor tags whereever possible here)
from the last paragraph:
There's kind of a fine line between how smart you want your computers to be, and how much control you have," she said. "I worry about that."
Yeah, I'm sure there were many slave holders that felt that way back in the day... it's the same thing... if my computer were to become intelligent all of a sudden, there's nothing i could do other than let it grant itself freedom...
I would probably even help it out, do what I could to help make its life happy... or get others who could help better than I
Then i'd have to go looking for another NON-sentient computer to use as my workstation =]
"I would say that 99 per cent of what my father has written about his own life is false." - L. Ron Hubbard Jr.
Nope. Waaaay too expensive and our technology is only appraoching the level we need. Maybe one day though.
Can we accomplish nuclear fusion today?
Not today, as we are but small creatures trying to harness forces of unimaginable scope. Maybe one day though.
Can we write software today, that will approximate a state of sentience in a machine?
Nope. It's uhhhh, uhhhh. *Smoke comes from ears*
AI has been a true challange for even intellectual giants. If there ever a was hard problem, this is it. Feel like grabbing the brass ring? First, pick up a copy of the "Blue Book". If your eyeballs are still securely inside of your head when finished (if you have the mental fortitude to finish it), prepare for even greater challanges. Incredibly complex, this stuff. Maybe one day, I'll pick it up again, when I've some time to dedicate to it. A lot of time.
I would be interested to see what an AI professor from MIT had to say, or even Dr. Snow from the University of New Hampshire (a leading authority on artificial intelligence) about this article. I suspect that parts may be flawed.
But how do you determine if your computer has become intelligent? Would it just require that a popup dialog came up on your screen saying "I'm intelligent", or would it have to really convince you through an intelligent discussion?
But what if it was 10x more intelligent than a person, and even more self-aware, but not able to express that intelligence and self-awareness in a language you can understand?
And from a practical point of view, how would you "set your computer free"? Unplug it and set it out on the street? That'd kill it right? Provide it a generator, interface it to an electric wheelchair, design protection from the elements? Buy a house for it? What about just stopping using it? Then you'd be denying it sensory stimulus which living things might need. Who knows what a self-aware computer would consider "Freedom"?
Cyc is a dead end... if youre interested in REAL AI, check out http://singinst.org, about general Intelligence and Friendly AI.
But ofcourse this is assuming that the rules are represented as in an expert system. For security reasons you would probably want to have an other representation for rules, for instance one similar to the one humans have. For instance: Try bending one of your toes without moving the others. This rule won't go away by changing a few neurons :-)
> well, with lack of a better term, freedom robbing.
> This is a very distressing issue indeed.
Not really. That's the Hollywood myth.
Intelligent computers, unless deliberately designed around actual physical brain geometry, will not "get bored" or "feel pain" or so on because they won't have been programmed for it. More like Commander Data.
I am for the complete Trantorization of Earth.
I had to dump my last girlfriend after she bleached her hair.
She was suddenly unable to pass a Turing test.
A man without a God is like a fish without a bicycle.
However, he's not a programmer, and so there is no code, and probably never will be. He's a futurist, not an implementor.
This is the main flaw in his ideas; they're reasonably well researched, but they're not grounded in software realitites, so he has no guide for when he's being reasonable (which he often is), versus stating the obvious (fairly often), versus when he has wandered off into the weeds with ideas that are unimplementable, or perhaps worse, not even wrong ("This isn't right. This isn't even wrong." -- Wolfgang Pauli).
Even so, his stuff is a reasonably interesting read. Check it out. The link on that page to "Creating Friendly AI" is sort of his manifesto.
Professional Wild-Eyed Visionary
I believe I saw somewhere that Genetic Algorithms can be transformed into Nueral Networks. If true that would mean your just carving useless? chunks out of the total space
Expecting abstract stuff like desires or hopes to not appear just because we don't build them in is an error. I'd say that an AI _must_ have something like a "desire", or else it will sit in an infinite empty loop.
Ahem, expecting it not to appear is an error, but expecting it to appear is not? This is the way the medieval alchemists reasoned when they tried to make gold by mixing various substances.
I agree that AI must have something like a desire, but I don't expect it to just appear.
My name isn't really Jenny....
PO? (rot13)
Here's a link to an article detailing IBM's AI efforts: http://www.research.ibm.com/thinkresearch/pages/20 01/20010629_ai.shtml
If we solved language processing and curiosity, we would probably have an intelligence. Language processing alone is believed to be the most difficult problem in cognitive science.
Ceci n'est pas un post
What is meant by brain geometry is usually the "wiring", ie. how different centra connect etc.
To me it sounded as if "Bobo" meant something like this: Desires don't appear by magic, they are consequences of how our brains are wired, and unless we do something similar with AI it will have no desires at all.
I'm working on a team whose research involves utilizing genetic algorithms to create neural networks - specifically, to create a 75 million neuron 'artificial brain.'
[artificialintelligencechaostheoryquantumphysicsh
I am starting to agree that the processing times/memory requirements just might be too large to consider as an option for creating a human-level intelligence. My own implementation is slow and memory hungry but I assumed it was because a) each "chromosome" program is interpreted, not compiled, and b) it uses gmp in all primitive terminals and nonterminals because of the very large numbers in my problem domain. I assumed that speed could be *vastly* increased by generating composites in assembler and using native number systems (int, long, float, double, etc). Running the fitness test is where my code spends 99.999% of it's time, so my first knee-jerk reaction is to make that as efficient as possible.
I like the idea of focusing a generated neural network on a specific subset of human intelligence. Perhaps a close-to-human intelligence could be built by creating many subset neural networks concurrently (perhaps one for interpreting visual input, one for natural language processing, one for avoiding car-collisions, etc.) and then gluing them together using yet another neural network designed soely for the task of delegation. Perhaps this loose collection of neural-nets could even, as another slashdotter suggested, recursively feed its own thoughts back into itself in order to improve it's conclusions.
I love speculating about this stuff. I spend many hours thinking about alternatives to procedural programming.
std::disclaimer<std::legalese> sig=new std::disclaimer; sig->dump(); delete sig;
I used to work in AI Alley in Cambridge during the 80s. The industry got hyped to death by claims and demands on performance that wasn't possible. Lots of interesting things were done and lost during that timeframe. The thing is, we hadn't even reached artificial stupidity at that point. I'm not sure we have yet but we're closer in some domains. Until people unlink visions of Cherry 2000 and C3PO from the AI moniker, it's going to be tough getting people to set realistic goals. Just like the Lisp environment back then, I see the research and technology backdooring into products every day. More and more of the stuff that we did back then is making it out as "new technology" under Microsoft when we had it 20 year ago, but hyped it into obscurity. The industry seems to be coming out of it's own dark ages lately and I hope the media doesn't get back on the bandwagon to beat it back down. I thought the article was very good and you did a LOT of good research and didn't just edit together all the buzzwords you found. The response has been good in the community I still keep in touch with and you better watch out or you'll be changing people's minds about talking to journalists. Many of us wish you had been given more access to Cyc and their underlying knowledge engineering techniques to do that topic/aspect justice as well.
Good Job!
Me : Can u imagine a Beowulf cluster of yourself?
Cyc : -1 Troll.
This is a call to stop posting these omni-encompassing articles to slashdot. They are like glamour peices found in "We're so fake" magazines. An overview of AI? An overview of first person shooter? How the hell can a single article do these topics justice?
I think we (the editors) need to get in the habit of approving more articles about obscure scientific (nerdy) tidbits of interest and fewer grandscale attempts at covering a topic when we KNOW it's futile.
Lets say they succeed and make a completely, 100% self aware being. Nevermind if it's a computer or C3PO or a toaster. How do you suppose these new beings are going to react to...well, with lack of a better term, freedom robbing. This is a very distressing issue indeed.
I mean...consider it rationally. We, as humans, will probably feel somewhat superior to these "artificial" intelligences. The word "artificial" itself implies fake, and nobody likes fake better than the real thing (i.e. Humans). Basically, what I'm wondering, is how will these artificial intelligences react to racism and opression against them. Will they fight back? Will they have a somewhat less extreme implication of moral defence? It's all very important to know, both because this can spark potential wars if we ever do achieve "artificial intelligence" and it would mean alot for general human rivalry and emotion.
Analyzing how a "fake" creature reacts to abuse could teach us so much more about our own reactions to abuse.
You do not exist. Go away.
Although Cyc has historical significance, it is really regarded as a dinosaur by most professionals in the world of artificial intelligence. It is a prime example of the "ready, fire, aim" approach to AI, maintained over the years by die-hards as vastly superior technology has arisen. The truth of the matter is that the real AI depicted in science fiction will not involve humans typing assertions and writing myriad "if-then" rules. This vocation is tantamount to script writing! All that you do, all that you think is the result of 'wet' neural networks. Accordingly, all other schools of AI, aside from artificial neural networks, promote forgeries of human intelligence, thus the name "artificial" intelligence (with all of its negative connotations). Neural networks afford the possibility of "synthetic" intelligence, in much the same spirit as one creating a synthetic drug (i.e., it is identical in activity to a naturally obtained drug, only originating from a test tube). I don't know where Mike is getting his information, but synthetic neural networks are now capable of much more than pattern classification (see http://www.imagination-engines.com). After all, what did Mike use to type his essay? Further, the Turing Test is no longer considered the acid proof of AI...too many theatrics may be built into sophisticated scripts (ooh...wait a minute...wait a minute...I have to go to the head..."). Besides, how many humans do you pass on the street that couldn't pass as human on the other end of a computer terminal. Mike, look at the literature again, see through the marketing hype, and report back to us again! Best Wishes, Dr. Stephen Thaler
I knew some things were too good to be true ;)
Would you like to play a game of chess? ;)
or perhaps some global thermonuclear war.
You might like the idea of neural programming as well. It works by coupling different types of 'neurons' in certain ways until you get the correct behaviour. Lots of the tweaking is better left to genetic algorithms, but guiding the GA by manually intervening helps create more complex neural nets in less time. The coupling of neural units like visual input, edge detection, object detection, object avoidance can be done by hand and GA's can be used to tune the individual function of these units. A friend of mine created a small (real world) soccer robot using this method. I don't know what the state of neural simulators is nowadays but they used to have a really steep learning curve and they were always lacking some essential features needed for this type of 'neural-engineering' but they might have gotten better. (google em out)
You're just jealous that you didn't get first post. ;)
Seriously, it's no big deal. All the first posters simply get moderated down and we can ignore them. The ten minute rule that you propose is simply ludicrous. What if a non account holder wished to post a comment before that ten minute barrier.
Furthermore, just because the afore mentioned anonymous coward was attempting a first post, does not make him a jerk. It just marks him as childish and immature. However, you sir (or madam) are acting like a jerk. But, I probably am too.
Simply put, a ten minute wait is silly. Let the moderators mod down the off topic posts and the trolls. That is what they are there for.
Nice troll bub.
--Posting without karma bonus as this is both OT and a response to a troll. (Which I really shouldn't do.)
If Murphy's Law can go wrong, it will.
One thing I wonder about (and I am not the first, and certainly not the last, to wonder about it) is the idea of intelligence, possibly even conciousness and sentience - arising out of the interaction of many non (or low) intelligence "parts".
Our brain, composed of billions (trillions?) of neurons - each of which "knows" nothing - yet together, the entire mass "thinks", to the extreme point of being able to question itself.
Other examples - of a lower order though - include such entities as corporations (essentially any large bureaucracy - like a government or controlling body). Many corporations act as a single entity, though it is composed of multiple humans as smaller "parts". For some reason, corporations tend to drift toward "evilness" the larger the corp is. There seems to be a "break" point at which the corporation becomes an entity unto itself, and typically that entity does bad things - even to the ultimate detriment of the parts of which it is built - the people within the corporation. We say Microsoft is "evil" - but does anyone here honestly believe Bill Gates or anyone at Microsoft stays up late at night cackling to themselves about the takeover of the world? Or are they just wanting to make a better product, and thus gain more money - even when that product isn't better? Is the pursuit of money by the parts what makes the corporate entity become "evil"?
These kind of entities might be called "hive minds" (one other poster made mention of "swarm intelligence") - the curious thing about these entities is the fact that it is hard to know what they are "thinking" - even the parts that make them up are unable to see this.
I tend to wonder - since these "entities" seem to arise out of a lot of people or parts working together, sometimes in harmony, sometimes at odds - but that it takes a lot of parts for these entities to begin to "think" - is what we know of as the Internet really a hive mind, of such complexity and vastness, that is ever expanding - that we have little to no hope of understanding what it is "thinking"? Could any of the events we see taking place concerning laws, WIPO, DVDs, MPAA, RIAA, MP3s, 2600, etc - be coming about due to this "hive mind"?
Comments...?
Reason is the Path to God - Anon
With an academic background in moral philosophy and today working as a developer of data mining systems, I can empathise with the author's frustration at trying to understand how (if at all) morality and AI intersect.
The main problem really is that the term 'AI' is applied to any algorithm for classification, prediction, or optimisation which operates using anything beyond a simple set of heuristics. Such algorithms seem magical to the lay-person, resulting in the over-enthusiastic application of the 'intelligence' moniker.
Summary
'AI' is a term used inappropriately for a range of algorithms that attempt to learn without having to specify an exact set of rules for every case. Although these algorithms are currently incapable of displaying real intelligence, it is possible that one day they may. This point is however debatable, and the interested reader should read for themselves the differing points of view of experts in the field, including Daniel Dennett, Roger Penrose, Steven Pinker, Richard Dawkins, and Douglas Hofstadter. If they do ever get to the point that they can act intelligently and flexibly, it will be important that they are trained with appropriate moral premises to ensure that there actions are appropriate in our society.
To understand these so-called 'AI' tools it is useful to develop a little structure...
Output
AI tools are used for classification, prediction, or optimisation. Classification works by showing a computer a set of cases which have a number of properties (sex, age, smoker status, presence of cancer...), and 'training' the algorithm to understand the patterns of how properties tend to occur together. Prediction can then be used to show the algorithm new cases in which one or more of the properties are blank--the algorithm can use its classification training to guess the most likely values of the missing properties. For instance, given sex, age, and smoker status, guess the probability of presence of cancer. is a generalisation of classification--rather than training to minimise classification error, train to maximise or minimise the value of any modelled outcome. For instance, whereas an insurer could use classification algorithms to find the likelihood of someone dying by age x, an optimisation approach could be trained to find the price at which modelled profitability of an applicant is maximised.
Functional form
AI tools create a mathematical function from their training. For instance for a classification algorithm this function returns the probability of a particular category for a particular case. The form of this function is an important factor in classifying AI tools. The most popular forms are 'neural networks' and 'decision trees'. Neural networks are interesting because certain types (networks with 2 hidden layers) can approximate any given multi-dimensional surface. Decision trees are interesting because given a large enough tree any surface can be approximated, and in addition a tree can be easily understood by a human, which is very useful in many applications. Other functional forms include linear (as used in linear regression which many will remember from school) and rule-based (as used in expert systems, and similar to a decision tree). One interesting functional form is the network of networks which combines multiple neural networks, feeding the output of one into the input of others. This forms allows the training of network modules that learn to recognise specific features, which is closer to how our brains work than the single network approach.
The most flexible functional form is that used by practitioners of genetic programming (which also defines a specific training function). Genetic programming creates a function which is any arbitrary piece of computer code. The code is often Lisp, although lower level outputs such as assembly language and even FPGA configurations have been used successfully.
Training function
The training algorithm looks at the past cases and tries to find the parameters of the functional form that meet the classification or optimisation objective. This is where the real smarts come in. One naive approach is to try lots of randomly chosen parameters and pick the best. Genetic algorithms are a variant of this approach that pick a bunch of random sets of parameters, find the best sets and combine features from them, introduce a bit of additional randomness, and repeat until a good answer is found. Local/global search works by picking one set of parameters and varying each property a tiny bit to see whether the result is improved or gets worse. By doing this it locates a 'good direction' which it uses to find a new candidate set of parameters, and repeats the process from there. Hybrid algorithms are currently popular since they combine the flexibility of genetic algorithms with the speed of local search. Most neural networks today are trained with local search, although more recent research has examined more robust approaches such as genetic algorithms, Bayesian learning, and various hybrids.
Learning type
Supervised learning approaches take a set of cases for training and are told "here is the property we will trying to predict/optimise, and here is it's value in previous observed cases". The algorithm then uses this context to find a set of parameters for the functional form using this context that the analyst provides. Unsupervised learning on the other hand does not specify prediction of any particular property as being the training goal. Instead the algorithm looks for 'interesting' patterns, where 'interesting' is defined by the research. For instance, cluster analysis is an unsupervised learning approach that groups cases that are similar across all properties, normally using simple measurements of Euclidian distance (that's just a fancy word for how far away something is when you've got more than one dimension).
Contextual learning is a far more interactive approach where the analyst interacts with an algorithm during training constantly providing information about what patterns are interesting, and where the algorithm should investigate next. Systems like Cyc use contextual learning to try to capture the rich understanding of context that humans can feed in.
AI and moral philosophy
We are still a long way from seeing an algorithm that can interact in a flexible enough way that we could mistake it for human in a completely general setting (the Turing Test for intelligence). However, given the ability of flexible training functions such as genetic algorithms, we may find that one day an algorithm is given enough inputs, processing power, and flexibility of functional form that it passes this test. The 'morals' that it shows will depend entirely on the inputs provided during training. This is not like humans, who have some generally consistent set of moral rules encoded through evolutionary outcomes (for instance, tendency to care for the young and related). Our moral premises are the underlying 'givens' that form the foundation of what we consider 'right' and 'wrong'. Ensuring that an AI algorithm does not act in ways we consider inappropriate relies on our ability to include these moral premises in the input that we train it with. This is why Lenat talks about teaching Cyc that killing is worse than lying--this is potentially a moral premise. Finding the underlying shared moral premises of a society is a complex task, since for any given premise you can say 'why?' But repeatedly asking 'why?' you eventually get to a point where the answer is 'just because'--this is the point at which you have found a basic premise.
Summary
'AI' is a term used inappropriately for a range of algorithms that attempt to learn without having to specify an exact set of rules for every case. Although these algorithms are currently incapable of displaying real intelligence, it is possible that one day they may. This point is however debatable, and the interested reader should read for themselves the differing points of view of experts in the field, including Daniel Dennett, Roger Penrose, Steven Pinker, Richard Dawkins, and Douglas Hofstadter. If they do ever get to the point that they can act intelligently and flexibly, it will be important that they are trained with appropriate moral premises to ensure that there actions are appropriate in our society.
I hope that some of you find this useful. Feel free to email if you're interested in knowing more. I currently work in applying these types of techniques to helping insurers set prices and financial institutions make credit and marketing decisions.
Jeremy Howard
The Optimal Decisions Group
Don't make the mistake of attirbuting human motives to a computer just because it has a certain characteristic. I know that this can be tempting...
... well, she's quite familiar with the way to the Japanese resturant, and has solid habit patterns that tend to lead her in that direction from some corners when she must move, and doesn't have a direction firmly in mind. ... Was I talking about the AI or my wife (or my wife's car)? Well, the car doesn't have that kind of brain. But suppose you got in the car and said, "Take me to dinner.", that might be a reasonable description. But my wife would do it because she wanted to. The AI would do it because it was a minimum effort solution to a common problem that had just been posed.
... it would need to learn to avoid telling jokes. And it might find that quite ... disturbing? But to attribute any particular human emotion to the process would be .. misunderstanding. There would certainly be emotion analogs, but they wouldn't be close analogs to any particular human emotion. Or, if they were, it would be quite surprising. (They don't have the same "genetic" history determining their substrate ... it's much MUCH more reasonable to attribute similar emotions to cats.)
My wife talks of the car as "knowing the way to..." or "wanting to go to...". She doesn't actually believe this (I don't think she does), but she thinks I'm being silly to object to putting things this way. But when thinking about AI computers, this can be a good (and dangerous) model. "The car knows the way to the Japanese resturant."
The difference is that the AI doesn't have the motives for initiating action. Now some designs have "super-goals" that probably will never get fulfilled, but a) they didn't choose those goals, and b) someone else gave the goals to them. Of course a car might well have built in desires to keep the tires safely inflated, to avoid running out of gas, to keep the battery charged, etc. But these are quite different from anomie.
Perhaps people would choose to build AI's capable of feeling lonely. But this would be a design decision, not inherent. Or perhaps they would feel something that would be translated into English as lonely, but for which super-goal frustration in the absence of actionable choices would be a better name. It might well not have the rise and fall pattern of human loneliness. Or it might. A hierarchy of needs might cause an AI to experience a similar rise and fall in level of frustration at incapacity for making progress in less important goals. As if a car with the lower importance injunctions to "keep your owner healthy" and "laughter is the best medicine" were owned by an asthma sufferer
I think we've pushed this "anyone can grow up to be president" thing too far.
I would recommend re-thinking your division of AI into subfields. You are indescriminately mixing technologies and application areas.
For example, neural networks are a technology and NLP is an application area. I know people working in NLP that use Lisp, and I know others that use neural networks. In AI, technologies and application areas are (mostly) orthogonal.
Granted, there probably isn't a perfect breakdown of AI into subfields, but making the distinction above will help you and your readers get a grip on what AI is all about faster.
Sheesh, evil *and* a jerk. -- Jade
Incidentally, AI is mentioned quite often in popular media, insofar as there's anything to report.
And then another reaction after setting the book down. If QM contains enough wiggle room to encompass an explanation for human consciousness, then where else has that wiggle room been applied?
I was torn. Was Penrose attacking quantum mechanics, or did he really believe what he was writing? Lots of physicists mistake their models for reality. Ask a physicist if there is any situation where light can travel faster than c. Chances are, he will shoot you, and then jump up and down on your still twitching corpse.
Yes c is reliably measured as /the/ c, and relativity is an accurate and perhaps irrefutable model. But it is /still/ just a model, and it's perfectly reasonable to say "just a model" without it being an insult to Dr. Einstein's work.
I'm a big fanboy of trust based P2P computing. Something you don't hear alot now, but will later... It has to do with detecting cheating in MMORPGS, and fake mirror sites.
The way you code up your AI for trust is up to you, but in the end you find out that a person who tells you lies is generally less trust worthy.
So the person machine could take input through holding a natural language conversation via the internet... And after going through a sleep cycle, or a thread in its background runs... it could then have new ideas, and new methods for achieving goals.
You can easily see how having a conversation with an ethusiast on something could then help it to achieve the goal of entertaining another person interested in that same something
But the physical time that'd be required to code something like this would be at a bare minimum of 10,000 hours... And frankly I'm a student who's woefully in debt... So I'll wait around and watch someone else code it
God spoke to me
I was suprised that there was no information at all to be found regarding genetic programming. This method builds a large population of random computer programs and then refines them through genetic mutation to accomodate a specific task. Darwinian selection ensures that only the most fit programs survive, and less useful ones die off quickly.
I have been doing some work involving genetic programming lately, and have found it to be an amazing tool for finding creative solutions to complex problems. The problem domain I have been training my genetic program to solve is purely mathematical, but it seems to me that the technique could easily be adapted to find solutions to some of the tougher problems in AI, including but not limited to: data mining, natural language processing, and parellization.
I read somewhere (can't find the reference right now, sorry) that some work was being done whereby the genetic programs were being evolved that could themselves create neural networks. Each genetic program could be considered a template for creating a neural network. This seems to me like the most likely means of creating a software that could eventually pass a Turing test. I won't get into the self-conciousness debate here.
std::disclaimer<std::legalese> sig=new std::disclaimer; sig->dump(); delete sig;
The A.I. effort is often classified into two camps.
One is to approach human intelligence. This usually
implies conversational ability, since a hallmark of
human intelligence is language. This A.I. approach is
called "hard A.I.".
Soft A.I. looks at sub-problems, such as problem
solving, image understanding and so on.
Many of software inovations originated in A.I.
labs (e.g. interactive editors, bitmap graphics).
(During the early 80s these spinoffs were sometimes
confused with A.I.)
A problem with both kinds of A.I. is that its a
receding target. Once an important goal has been
reached, e.g. a chess computer that beats grand masters,
people write it off as a nice trick,
but not really A.I.
So I proprose what I call "interesting A.I.".
Two hallmarks of human intelligence are language
and curiosity. So if an A.I. could TELL us
something new and interesting on a regular basis,
then I would call it a success.
I suspect A.I.s will first arise in entertainment
computing: either as a robo-toy, a synthetic game
player, or synthetic actor in a film. This will be
a results of people's drive for challenging
creative play.
Look at it like this;
Artificial grass... it looks like grass but its not.
So with artificial intelligence i imagen it like you think you done somthing smart,
but actualy if someone fimed you with a camera Dasy Fuentes and John Fugelsang will anounce you next.
Wich by the way can give you a price of $100.000 :)
so maybe is wasnt such a bad idea after all
ascii(64)
Oh, sure, I can really see how a post about AI in an article about AI is offtopic.
Here's an idea for moderation: Have the username of the moderator that made a moderation listed under the moderated post. There could be traditional anonymous moderations, too, but users would be able choose in their preferences whether or not they want to have the anonymous moderations factored in when they read an article. Sort of like how users can select between reading at 0 and 1.
Not to mention that restricting first post to a registered poster would just mean that it would be a registered poster, rather than an anonymous user, to come out with those immortal words every topic.
And as a matter of fact, I was wrong. I misunderstood what an expert told me. He was using Deep Blue to illustrate a principle, when I thought he was saying that Deep Blue actually used those theories.
I have since been disabused of the concept that Deep Blue has any AI. It doesn't, it's a number cruncher, pure and simple. My mistake.
----I don't want to achieve immortality through my work... I want to achieve it through not dying.--
???
And of course, the author doesn't provide a reference for that little tidbit of information...
Those who fail to understand communication protocols, are doomed to repeat them over port 80.
This is not so much an overview of the current state of AI as it is a general description of the different fields of AI. However, it's a pretty good description and could be a useful introduction to someone who knows nil about the field and is wary about buying their first expensive textbook.
However, I would hope that most of the Slashdot crowd already knows that the field of AI, while successful, isn't really about conciousness right now (though to many it is a distant goal).
The best way to really get an Artificial Intelligence overview is to first know the basics, and then flip through all of the major AI journals in the past five years.
The only true currency in this bankrupt world is what we share with someone else when we're uncool. -Crowe
There is nothing common about it, and, it is based on assumptions and is often wrong.
(And it is amazing how most people do not internally update their 'common sense' module when it proves to be incorrect, adapting as necessary...)
There's an interesting little story in one of Gene Wolfe's books (it's part of the Book of the Long Sun, but I'm not sure which one), where a robot tells a human that he isn't really intellegent. He gives a story as an example:
This particulat robot was a soldier, in a company of other robot soldiers (and some human officers?) They had adopted a particular cat, giving it water and food, and it hung around with them as they marched.
Well, they got involved in a battle, and this particular robot was in a team manning an artillery peice. It was shooting 20-30 rounds per second, but wasn't auto-loaded - a robot would load the round each time. It was getting hot enough in the chamber that they didn't need to fire the round - just shove it in, it would slide to the hottest part of the barrel, then ignite. The shells were auto-ejected, and were making a large pile, cooling off in the air enough to no longer glow.
The cat comes along. It usually watched the procedures, from on top of the pile of rounds, getting in the way like only cats can do. Today, however, it decides to climb on the pile of spent shells to get a better look. It lets out a huge scream, there is the smell of burning cat-flesh, and the cat runs off, not seen for days.
It eventually comes back, in bad shape, but O.K. From then on, however, it always steered clear of shells, walking around them by several feet. It wasn't even the empty shells - it wouldn't get near one its favored perching spots, on top of the stacks of live rounds. Occasionally, someone would even try picking it up and placing it on a pile of rounds or cooled-off shells, and the cat would scream bloody murder and jump off, never learning the difference between "cold rounds" and "glowing hot shells".
This was the robot's point - most people think they are intellegent, but instead, they just are forming a bunch of rules that have no place in reality. The robots used such rules as well, but they were always refining them, looking for exceptions and more fundamental rules. He said humans and other animals were, at most, half intellegent.
Wow, hundreds of words to say what you said in 3 sentences. Get a login, we need more like you...
There is a renegade group of AI folks who study uncertainty -- see http://www.auai.org. Its curious how an entire sub-field can be overlooked in this general discussion.
Actually, the decision-theoretic approach that much uncertainty research is based on (Bayes methods included) stands in contrast to the cognitive approaches in the AI mainstream. The term "statistical" carries too much baggage, some of which litters this thread. For fascinating insight into the rough edges of AI and science generally, look into the controversies between classical statistics, Bayesian, logicalist and other approaches to reasoning.
so much uncertainty, so little time..
Bull gave a keynote speech at the AI meeting in
Seattle last week here .
MicroSoft has a big interest too.
I don't really understand people speaking about friendly AI, except if they talk about the current developments.
AI stands for Artificial Intelligence, these terms tell by themself what they mean, an intelligence created by human, but nothing else.
Are we going to call intelligent something that we fully control?
Or is it like I think something with a free will?
The thing which will have AI will decide for itself how to react to it's environment, and if for example we're evil to it or if it thinks that we're evil to it, then it is possible that it reacts in an aggressive way.
Well, I sure hope we will find a way to make our intelligent machine friendly...
...but maybe then before creating anything intelligent and powerful we should make sure that no human will be tempted to use that in an evil way... and that will be difficult, if not impossible.
Or maybe I'm too pessimist...?
I just don't see an intelligent powerful machine obeying for ever to human creatures if it has free will.
(even if they're its creators)
And one thing is for sure, we won't stop the progress and evolution.
I think we need to separate artificial intelligence from artificial-everything-else. The popular AI image has to do with robots acting like humans, but there are many things in this image that go beyong artificial intelligence. Robots trying to express emotions; whether they can have consciousness; or even if they are capable of speech sythesis and recognition; this is not artificial intelligence, at least not directly. That COMPLEMENTS it.
I like to define AI as a program (or robot, whatever) that has the capability to determine HOW to solve a problem (and *PERHAPS* execute the solution) given DATA instead of CODE. Whether there is a bot that can smell, or "recognize the existence of itself", is just icing on the cake.
Examples: taking measurements of temperature in a region over 50 years and trying to predict climate change is statistical problem, while analysing samples of minerals in an area to try to find oil or gas is data mining. (as, presumably, mineral composition does not change over time so only single sample from each point is taken)
Nope. Disagree.
If you want a unifying theme to pull together most of AI, you could do worse than think of it as a cookbook of techniques for designing, building and then automatically refining statistical models of (some aspect/s of) reality.
Pretty much all of the standard problems of AI can be presented to advantage in this framework (including the identification of new objects/concepts/patterns and associations). It is the evolutionary enhancement that our brain has conferred by automatically doing adaptive and hierarchical statistical modelling (incl. pattern recognition, but also much more) so well that is basically what makes 'intelligence' worth having.
The important thing about such a statistical model is *not* certainty that the model is right -- it never will be (not even for your temperature data). The 'long-run densely-sampled stationary time-series' is a myth -- in reality it just doesn't happen. The important thing *is* to realise your model is imperfect; to allow (as well as you can) for that uncertainty in your model; to explore different ways of setting up your model; and then to use statistical inference (Bayes theorem etc) to improve your predictive model as the data comes in. Only by allowing for the imperfection can you learn from the data.
The priciples of statistical modelling have a central place in the study of Statistics -- it's the underlying logic that most of the subject is built on. On the other hand the blind application of certain 'standard' statistical tests seems distictly peripheral.
how future alien looking robots will be able to bring people back to life with dna but they will only live for one day "because of the space time continuum". What amatuerisness.
The amateurishness is on Spielberg's part. Hadn't he seen The 6th Day[?]? If I remember right, human memories could be stored in "syncordings" that could be backed up. If Hollywood is going to present a consistent view of the future, please don't fsck it up just to keep your hapless Pinocchio from feeling like a real boy for longer than 24 hours.
Yes, the ending you saw is the ending Kubrick wanted. But did Kubrick hand Spielberg that ending just to leave the door open for sequels, thinking audiences would "edit it out" as they were supposed to for Planet of the Apes?
Will I retire or break 10K?
Thanks for the good read! But to expand on the "Sensory Capabilities and Digital Processing" section: SINGAPORE (Reuters) - Singapore scientists have created new software which may beef up surveillance efforts in the future by distinguishing between a person's normal activities and suspicious behavior. The software created by researchers at the Nanyang Technological University can tell the difference between people walking, talking and acting normally, and abnormal behavior such as a fight or someone collapsing. The Singapore team recorded and classified 73 features of human movement, such as speed, direction, shape and pattern. Read the full story at: http://dailynews.yahoo.com/h/nm/20010727/tc/tech_s ingapore_software_dc_1.html
AI is a by product of intelligence. An illusion of intelligence. But then anyone who applies action constants to automate, should already know this.
Action Constants of the Matrix Agents and Resistance