Company Claims Development of True AI

YF 19 AVF wrote to mention a press release on Yahoo from company GTX Global. They think they've got a good thing on their hands, going so far as to claim they've developed the first 'true' AI. From the release: "GTX Global Cognitive Robotics(TM) is an integrated software solution that mimics human behavior including a dialogue oriented knowledge database that contains static and dynamic data relating to human scenarios. The knowledge further includes translation, processing and analysis components that are responsible for processing of vocal and/or textual and/or video input, extracts emotional characteristics of the input and produces instructions on how to respond to the customer with the appropriate substantive response and emotion based on relevant information found in the knowledge base." Somehow I think there is a littler hyperbole here. In your estimation, how close are we to the real thing?

Comments

How about

[Anonymous Coward]

not even close?

Re:How about

[Anonymous Coward]

First 'true' post

Re:How about

[SerpentMage]

I am not so sure.... BUT there is one easy way to tell... Put it to the Turing test. http://en.wikipedia.org/wiki/Turing_test. If it passes that, then I will willing to call it the first AI...

Re:How about

[MaskedSlacker]

Hell, Eliza passed the Turing test. Just ask all the people that felt they had successful therapy sessions with it.

Re:How about

[MrNougat]

The Turing Test presumes that the tester is an adequate judge of what is a cognitive response. However, if the tester is an idiot, the threshhold for what qualifies as a cognitive response drops.

It's clear to me that the Turing Test, while interesting, can easily be more a reflection of the tester than anything. A good IQ test, for example, would be to construct an AI that was at a benchmark level of intelligence. A person taking this IQ test would interact with the AI and a "referee" would observe. It would be obvious in short order which of the two was smarter. A series of AIs at varying levels of intelligence would act as benchmarks on a scoring scale, so a battery of such IQ tests would produce a standard IQ score.

Re:How about

[Doomstalk]

A true turing test requires you to talk about a wide variety of sibjects, whereas ELIZA forces you into a faux psychotherapy session. It's not the same.

Re:How about

[MasamuneXGP]

Turing test turing test... Psh. There are better and more accurate ways. Put the thing in front a Go board and we'll see if it's a capable AI or not.

Re:How about

[Anonymous Coward]

Whatabout making it some mechanicle hands and giving it a rubix cube?

Re:How about

[Kingrames]

Are you happy about "Eliza passed the Turing test."?

Re:How about

[jolyonr]

It's not that artificial intelligence is getting smarter, it's that people are getting more stupid, that's how the gap is closing.

Jolyon

Re:How about

[ephex]

Right. We'll never develop true AI until we completely understand the brain on every level, including molecular. As it is now, AI is essentially a misspelling of heuristic algorithms.

Re:How about

[it_ain't_my_fault]

Just tell the machine a paradox. I am lying! Is is true? and watch it explode.

Re:How about

[digitalunity]

Nah, just the opposite. If you tell it a paradox and it explodes, it wasn't capable of coming to a logical conclusion but couldn't understand that no logic can come of paradoxes. If I tell my computer a paradox and it tells me I'm illogical and that paradoxical statements cannot be computed, then I will believe it is the real thing.

Re:How about

[sgt_doom]

You know, I recall something like this back in the late '70s or early '80s when some clown nobody had ever heard of before, was supposed to be some kind of genius because some stupid, ignorant newsy with The Washington Post newspaper proclaimed him as such, predicted he was going to create the most advanced information-gathering computer system mankind had every seen.

If some super-technoid genius exists who is over the age of 18, many of us will have already heard of him or her.

Obviously, this ain't the one -- not even by a long shot.....

Anybody note this issue of Scientific American rebukes that silly urban legend that we typically only use 10% of our brains??? That fiction also derived from another stupid newsy back in the '40s; probably the father or grandfather of that Washington Post clown. (Anyway, I used at least 11% of my brain.)

Re:How about

[VernonNemitz]

If it doesn't have free will, then it isn't intelligence as we know it.
If it does have free will, then why is it so cooperative as the press release brags about?

Move Along.. No Marketing Hype to See Here....

[fyrie]

LOL

True?

[Seumas]

If it's true AI why does it just "mimic"? Isn't that what CURRENT AI does?

Re:True?

[Anonymous Coward]

no, "current" AI mearly is a preprogramed set of resposes, or a preprogramed algerithm to get a response.

one that mimics is a preprogramed set of... oh, wait .. nevermind

Re:True?

[Anonymous Coward]

"If it's true AI why does it just "mimic"? Isn't that what CURRENT AI does?"

Isn't that how people learn, by mimicing others. Of course people also eventually realize they have the ability to be original and develop their own characteristics, so right now AI does at least half the job. How much would be involved in giving it the other half? Maybe it's as simple as a good way of learning through random actions and observations.

Either way I'd rather have a dumb obedient computer, as soon as the things learn how to say "no" I think people will realize the mistake in making them smart.

Re:True?

[NitsujTPU]

Uhmm, that's just patently incorrect. You're thinking of expert systems, which represent a distinct sub-category of artificial intelligence.

However, if I go through this board correcting everyone, I'll never finish my final projects for the semester, so, I'll do it here.

Modern AI consists of a number of subdisciplines, each of which focuses on different things. I haven't RTFA, because I'll believe it when I see a paper on what they're doing.

To be brief, however, there are
Logical and Constraint programming, which focuses on solving problems through sets of contraints.
Knowledge Representation, which focuses on how we represent the world to algorithms that work on that knowledge
Natural Language processing, which deals with working with spoken language. It's considered that work in this field represents some of the hardest challenges in AI.
Machine Learning, which has been described as statistics on steroids by one of it's popular researchers when addressing his class
Human-Competitive/Human-Like AI, which generally works in bringing together these systems into a human-like intelligence
Multi-Agent systems focuses on behaviors of more than one agent
And others (I hope none of my collegues are offended that I didn't stick theirs in the list, but some of the descriptions get a bit intense, and, again, I need to get back to work)

Then you have all sorts of tasks:
Autonomous navigation
Word sense disambiguation
Game playing
Temporal and spatial reasoning
Planning
Scheduling
Tabletop space problems (which most closely resemble your "true" AI, and do not merely mimic the actions of the teacher)

and man others, again, I hope I've offended no-one, these are the ones in my head due to PhD apps being around the corner.

The president of the AAAI this year, called for what are called "AI Decathalons," whereby researchers would construct systems that do multiple tasks. For example, a system might take a written or multiple choice exam, which requires forms of reasoning, it requires naturual language to read the questions, it requires knowledge representation to represent the questions and data.

At the same conference, Marvin Minsky had remarks more of the flavor of "AI needs to change directions (dramatically)," but he still wouldn't constrain the accomplishments of modern AI to expert systems. His book "The Society of Mind," is probably not a bad place to start if you want to learn about modern AI. It's very accessible to people who have only a passing interest in the field, while having enough solid content, ideas, and commentary (from Minsky of all people) to keep a fairly advanced researcher interested. Also, if it comes up in conversation, it's one of those "it's good to have read" books, even if you disagree with Minksky's ideas (one such controversial idea, consciousness does not exist).

Re:True?

[mkavanagh2]

imagine a beowulf cluster of these things

Hot grits.

Re:True?

[DrSkwid]

The I part means knowing what to mimic.

Re:True?

[Oligonicella]

Being pedantic, your parent isn't entirely wrong. Unless there's some work being done on self design and modification in the sense that we ourselves change our behavior patterns. Seems to me that, until then, regardless of how very well you manipulate data, it's still just coding -- veerrry complex coding.

Your list tells me the subject is still in the stage of creating organism level responses to the world's stimuli, which then tells me that we're still at the beginning.

Re:True?

[NitsujTPU]

Some have said this.

Still, AI isn't just about learning patterns. There's an inductive step and varieties that are also about problem solving.

Most CP style systems and solvers definately don't imitate people, to say the least!

Re:True?

[bcmm]

Isn't that what WE do?

Re:True?

[Seumas]

Based on the information in your post, it would support my comment that all modern AI does is "mimic". If this company had invented "true artificial intelligence", it wouldn't be mimicing humans - it would actually be behaving like them. The old "AI" program that guesses what animal you're thinking of based on your responses (and later - contribution to it's database if it was something it didn't know before) is a great example of no intelligence - but mimicing human intelligence in a way that some people might perceive as actual intelligence.

Granted, "AI" includes the word "ARTIFICIAL", but I don't think that "artificial" means "mimic" and that is the wording that the company used in their release. Essentially undermines their whole "true AI" claim.

Re:True?

[NitsujTPU]

There's really far more to it than that.

No, you won't walk up to a robot and have a human-like conversation with it.

Yes, progress has been made.

Also, as I said to another poster, solvers and provers and constraint programs are nothing akin to human intelligence, nor do they try to be. That said, they are quite useful.

AI is entering our daily lives. Just watch.

Re:True?

[NitsujTPU]

Also, to clarify, there was a post between yours and mine, which was the one that made the patently false statement.

Yours has subtle shades of grey to it.

Re:True?

[Seumas]

I'm not so sure of that. Computers making choices based on structured rulesets are certainly entering our daily lives, but I think it will be a long time before something that does more than simply run down some sort of B-Tree of choices based on input. I was going to say that until an ATM can not only deny me because my PIN number is wrong, but because it "thinks I seem suspicious", I won't be convinced -- but even then, that doesn't prove any sort of AI as the computer will still be following a ruleset developed and programmed by someone with strict indications of what is or isn't suspicious.

I think there's a big difference between "artificial intelligence" and "intelligent emulation" and all we're getting so far is "intelligent emulation".

Also - do you really think that "expert systems" (if I understand them correctly) could ever be where the spark of true artificial intelligence occurs? I mean, by definition, wouldn't artificial intelligence occur in a less structured arena where they are not confined to one simple set of tasks and rules? If all your system knows about is watching for computer virus activity, generating signatures and relaying those signatures to sentries throughout the internet -- could it ever have anything more than "emulative intelligence"?

Re:True?

[NitsujTPU]

Machine learning is being used in the diagnosis of hospital patients.

Data mining is being done with machine learning techniques as well.

Space rovers use robotic motion planning algorithms, and soon will use algorithms to determine objectives (these aren't expert system type algorithms).

The cross-action toothbrush was developed using a neural network.

These are what I mean be entering our lives.

Expert systems had their hayday when big investment banks were using them, but interest in them has kind of burned out. There aren't any conferences on them or the like. Neural nets made a resurgence, as did decision trees in the past few years, so, who knows? I don't really think that expert systems will make that kind of comeback any time soon though.

I have to be honest in my analysis though. I h

Re:True?

[arose]

Computers should say "no" sometimes.

/home/user$ rm -rf * ~
AI: Maybe I should just delete the backups?

Re:True?

[CaptDeuce]

Then you have all sorts of tasks: ... Word sense disambiguation...

Disambiguation? Couldn't that refer to a number of different things?

:-j

Re:True?

[Redwin]

Technically speaking you just mimic as well, for your initial method of learning at least, so as a starting point it is to expected that AI gives the impression of mimicing others. Whether they can reach conclusions on their own with the information they already have is more interesting.

Re:True?

[arkanes]

If you continue your train of thought, you will come to the conclusion that human intelligence is just as non-existent as AI. When a teller thinks you're suspicious, he or she is applying a series of rules to it's input. Most of them are even documented, formalized rules that they were taught by instructors.

SKYNET IS COMING!

[Anonymous Coward]

BEWARE! SKYNET IS COMING! WE MUST STOP THIS BEFORE IT'S TOO LATE!

Lameness filter encountered. Post aborted!
Reason: Don't use so many caps. It's like YELLING.

True AI

[m0rph3us0]

When you develop "true AI" you dont make a press release about it, you phone the military of your country of choosing and wait for men to arrive with large briefcases full of money. Let me put it this way, true AI is not annouced by /., you will read about it in Janes about 10 years after it happens.

Re:True AI

[d_strand]

When you develop "true AI" you dont make a press release about it, you phone the military of your country of choosing and wait for men to arrive with large briefcases full of money. Let me put it this way, true AI is not annouced by /., you will read about it in Janes about 10 years after it happens.

Not if I invent it man. I hate the military and all they've done to us. If I invent true AI you can expect to read about it instantly... maybe even on slashdot so stay tuned! =)

Re:True AI

[DigiShaman]

Just to make AI military applications clear, you will never see such a program acting as a substitute for human beings. At best, it will be a supplement. Because of this inherent nature in the art of war, true independent thinking and self aware will rarely be given a chance in the theater of battle. Why you ask? Simple. It's a security risk!

Re:True AI

[jcr]

If I invent true AI you can expect to read about it instantly...

Why would true AI cooperate with you?

-jcr

Re:True AI

[Breakfast Pants]

"True independent thinking and self aware will rarely be given a chance in the theater of battle," if that were the case then currently only computers would be allowed in battle and humans would be on the sidelines, being that humans exhibit true independent thinking and self awareness (sometimes).

Re:True AI

[dtfinch]

If you don't want to sell it, they'll just take it by eminent domain. Then they could even classify it as a military secret so that it'd be treason to discuss your invention with others.

Re:True AI

[Anonymous Coward]

Actually, the military does like independant thinkers, to a point. The ability of lower ranking officers and NCO's to make decisions in the battlefield is very important to the United States military strategy. We are speaking in tactical terms, not strategic. A captain or majors ability to make tactical decisions in the midst of battle is a great advantage.

The Soviets had a military command style of the type your describing. The smallest decisions, outside of SOP, had to be bumped up the line to an extreamly high level, taking a large amount of time.

Re:True AI

[91degrees]

Maybe they asked the AI, and it had ethical objections ;)

Re:True AI

[Omestes]

Completely missed the point.

Why does the military brainwash soldiers? Simple, to render them compliant, and no free thinking. "Just following orders" is the goal, sad to say. This might not be true of officers and specialists this is less true, but for your average grunt, then yes it is ideal to be nonthinking.

Do you think bootcamp exists only to bread skill? That is what the schooling afterwards is for.

Same thing with police forces having IQ caps, you don't want people to question their job.

Re:True AI

[DigiShaman]

I have friends who've fought in the first Gulf War. Every one of them well tell you that "trust" of another fellow soldier is more important than the weapons they wield. So unless self aware AI can legally be held accountable for it's own actions for the decisions it makes on its own, I don't see such thinking machinery out on the battle field making such crucial political decisions (war is political).

Put it to you this way. Would you rather have an android with very little real-world experience or another human being fighting side-by-side with you? I'm sorry, but I don't want some software bug or glitch to slug me with a 50 cal round. To fucking risky IMHO.

Re:True AI

[lxs]

I for one can't wait to subscribe to Jane's Thinking Bots

Re:True AI

[Anonymous Coward]

Oh no! That's the first test for "true" AI. If given the choice of going to the military and going to the stock exchange it chooses military you know it is a fake.

Re:True AI

[Jekler]

Exactly. "True AI" is a far cry from a customer service system that assists in formulating appropriate responses. The applications of "True AI" are so vast as to be unimaginable. Games, military, production... a system capable of understanding is a tremendous accomplishment. At a minimum, an AI system must be capable of crafting solutions to situations it wasn't specifically designed for.

Re:True AI

[KDR_11k]

I'd prefer the android to fight INSTEAD of me.

Re:True AI

[Anonymous Coward]

haven't we already concluded that you can't ED intellectual property?

Re:True AI

[KDR_11k]

What would you say on the phone? "How are you gentlemen, my super AI has taken control of all your computers and installations. You are on your way to destruction if you don't pay me ONE MILLION DOLLARS!"

Re:True AI

[NitsujTPU]

We've already got those in the pipes. Read up on what the DARPA Grand Challenge is all about.

Re:True AI

[TapeCutter]

I agree it can't replace the military mind. If you let the a "true AI" see the big picture it would tell the generals to put the guns down.

Re:True AI

[DigiShaman]

You talk as though "true AI" would promote peace, but the exact opposite could be true as well. Maybe, they might strategically pin the human race against each other so they "AI Machines" can be the remaining victor to inherit the Earth. Insidious, I know. But just as equally plausible.

Re:True AI

[EtherealStrife]

You'd prefer a human being slugging you with a 50 cal round?

Atleast with the combat robots you can deactivate them after the fighting is over. With humans, it's either lock em up for life or attempt to reintegrate them into society. I'd rather have a machine sitting deactivated in a warehouse somewhere than a biological killing machine living nextdoor.

Re:True AI

[Aphrika]

I'd assume that when you develop "true AI", it tells you it's going to make a press release.

I'd also expect it to be involved in negotiations with bidders. However as this is just a database with "dynamic and static data" based on human scenarios, and it runs on bog standard computers, I don't see exactly how it can be construed as AI - it has no random element nor cognitive ability to think for itself outside of what it's told in its scenarios.

Re:True AI

[CarpetShark]

Sure, if you've no morality whatsoever. People who develop revoluntionary ideas like this are as likely to have altruistic goals as militaristic ones, though, if not more likely.

Re:True AI

[munch117]

Why would true AI cooperate with you?

Because an AI isn't necessarily motivated by self-perpetuation. Life forms developed through evolution are so motivated, but there's no reason to expect something intelligently designed to have the same motivation.

If I was designing an AI, I would make sure it understood it's purpose in life was to get me laid...

Re:True AI

[mkavanagh2]

More likely you'll read about it in Janes' 10 years before it happens.

Re:True AI

[Anonymous Coward]

Unless it's "evil" and/or has other ulterior motives.

Re:True AI

[DrSkwid]

And what a great job the US system does of ground wars.

Re:True AI

[Hrvat]

Actually, having been in the military and having IQ over the police "cap" I can tell you that the REASON for having someone in the army to follow orders is because most of the time the soldier does not have the complete view of the situation. You depend on the CO to have that information and make decisions accordingly (because there is often no time to do more than that). That said, soldiers of today are more independent than ever seeing how they have to deal with local populace without immediate contact with the CO.

Regarding the police, the IQ cap is there merely to prevent people getting bored with their jobs, because it takes a special kind of intelligent person able to deal with sitting around most of the day and filing paperwork. Most of police work is BORING.

Re:True AI

[jcr]

If I was designing an AI, I would make sure it understood it's purpose in life was to get me laid...

Ergo, no free will.

-jcr

Re:True AI

[igotmybfg]

What the previous poster means is that it is possible a software glitch would cause an android/robot/whatever to shoot him in the back. He estimates it is less likely that a fellow human soldier would do this. I agree.

Re:True AI

[moonbender]

Why would you create an AI that's suicidal by design? ;)

Re:True AI

[munch117]

If I was designing an AI, I would make sure it understood it's purpose in life was to get me laid...

Ergo, no free will.

It would have just as much free will as you and I. We have innate desires that we are born with and can't change. We desire to stay alive and procreate, to avoid pain and achieve pleasure. Sometimes we are under the illusion that we do things that we don't desire, but really that just means different desires are in conflict.

It's no different for the AI. It is born with whatever desires we choose to program into it, and has to live with that. We could choose to program an AI to desire self-perpetuation and procreation, but it would have no more free will for that.

Re:True AI

[spacecowboy420]

As a veteran, I resent your "brainwashing" assertion. Maybe to some it has that effect, but "brainwashing" in the military is no different than the "brainwashing" in any training. The fact is, in the military the stakes of your training are much higher - you are training people to deal with real life or death situations. The emperative is on accomplishing the mission, much like anything else you could learn to do, the difference is you could die if you fail in the military. Yes, this requires a certain amount of faith in leadership, but that is the nature of the military.

Re:True AI

[Jamesie]

Do you think bootcamp exists only to bread skill?

In the catering corp maybe.

Re:True AI

[Rakishi]

Not really, depends on how it's designed/trained. In the end if it's more intelligent than any human we simply can't comprehend much less predict its actions in anything but the most abstract sense.

Keep in mind that humans have certain inherent and social ways of thinking in them, which an AI need not have at all. Human life could be meaningless to it, and only the success of the "nation" is of any value.

As one of Asimov's short stories mentions the AI may decide than an agrarian society is the best for humanity, although lacking Asimov's three laws it may decide a well planned war is the fastest means to that end.

As for putting guns down in a more common sense, would it really? That just means the other side will decimate you, which may be much worse than the cost of the war.

Re:True AI

[rob13572468]

you're about an order of magnitude off if you have control of all computers. thinks BILLIONS. of course if it were me i would simply do it to provoke a showdown with superman...

Re:True AI

[patio11]

Haha, thats funny. I'm an AI researcher and have worked on, well, call it a related field with a related government agency. You think the DOD would actually need or desire "self aware" for any application? Or one of the generalized Data-type "its just like a human, except it has no physical brain" sci-fi AIs? Heck no. They'd want an algorithm which was the electronic equivalent of a blood hound -- doing one thing, very very well. The Holy Grail of military-application AI would be Google Search raised to the nth power -- something that could take raw, unprocessed data in an arbitrary format (e.g. here's a list of all the international bank transfers coming from Europe in the last six weeks) and exectute arbitrary queries on the data ("Bloodhound, we think there is a terrorist ring compromised of about twelve to twenty Muslim professionals with connections in Bonne, known sitings in Paris at the riots, and they're partly financed by someone with shadowy connections to the Saudi royal family. GO GET HIM, BOY!" and then, two hours later Bloodhound would say "The following 423 bank transfers are consistent with the supplied hypothesis. The cell's main locus of operations appears to be Lisbon. Analysis indicates that the Saudi connection is unlikely; the main source identifiable source of funding seems to be an Oil-For-Food slushfund which the UN monitors have missed." (It should be pointed out that this example is pretty darn sci-fi itself, but it is a heck of a lot more plausible sci-fi than any "self-aware" BS.)

Another potential field would be simple image processing. "Is that smudge a tank or a school bus?" Neural net spits out "School bus, p=.62, tank, p=.23, 1996 Mazda, p=.04"

Re:True AI

[Anonymous Coward]

Because an AI isn't necessarily motivated by self-perpetuation. Life forms developed through evolution are so motivated, but there's no reason to expect something intelligently designed to have the same motivation.

You have just proved the case for the intelligent design folks.
Humans can be seen everyday driving huge SUV's faster than the speed limit while chaatting on a cell phone and lighting up another cigarette on the way to Micky D's. Obviously, they aren't motivated by self-perpetuation.

Re:True AI

[RhettLivingston]

It is unlikely you'll read about it in Jane's 10 years after. 18 months after, it's intelligence will be doubled. 18 months after that, again. That may be the last 18 month long generation. It will quickly be intelligent enough to contribute to the design of it's next generation and start rapidly accelerating Moore's law. i.e. a true AI, initially created through accurate and faithful reverse engineering of the human brain, will be able to bootstrap its way through an ever accelerating evolutionary curve. In ten years time, the world will be changed in ways truly beyond our imagination because the intelligence making the change will be beyond our comprehension.

Re:True AI

[javachip]

ED is the wrong approach. National Security would have a lot more bite in this situation.

Re:True AI

[mre5565]

Just to make AI military applications clear, you will never see such a program acting as a substitute for human beings. At best, it will be a supplement. Because of this inherent nature in the art of war, true independent thinking and self aware will rarely be given a chance in the theater of battle. Why you ask? Simple. It's a security risk!

A human being in uniform is far more of a security risk than a military 'bot with AI. Just look at the number of court martials from the war in Iraq. Humans make mistakes. They humiliate or at worst torture human PoWs and then the photos get shown on CNN. Bots will most likely kill the enemy before he has a chance to surrender. CNN doesn't seem to mind if the enemy is efficiently and ruthlessly killed.

A military bot that can react as fast or better than a human soldier will pretty much do to infantry warfare that iron clad ships did to naval warfare. In the case of the latter, once it was clear an iron clad was feasible, all the wooden ships were instantly obsolete. So it will be with the human infantryman when the military bot is introduced.

Imagine the current Iraq conflict if the side with the all the money had military bots. No coalition or Iraqi gov't soldiers die, whereas the insurgents die at an even higher rate. Iraqi government officials simply conduct their business via telepresence in similar bots. Suicide bombing of police, soliders, and government officials becomes useless: "Today a Belgian suicide bomber rammed her car into an Iraqi police station and destroyed 3 bots" just doesn't make page 1 of the newspaper. Bring it on.

No substitute for human beings? In case you haven't noticed, a solid majority of Americans believes 2000 dead soliders in Iraq is about 1900 or so too many. If and when this technology becomes viable, it won't matter what people like you or in the Pentagon say. The public that pays taxes and supplies cannon fodder will demand military bots to replace humans in direct hand to hand combat. Sorry.

And if the US is foolish enough to not deploy the technology, eventually China will (though it will be much later as a Chinese human solider costs much much less than an American human solider, and there are far more Chinese humans), and then at that point any Pentagon luddites still left will be irrelevant.

But, given the trend of the current US military thinking, I suspect Pentagon generals appreciate the benefits of an independent military bot far more than you or I.

Re:True AI

[geminidomino]

Someone hasn't seen Austin Powers. ;)

Explained: The "One MILLION Dollars" is a running joke when the 70s-era "Supercriminal" emerges into the end of the 20th Century. He demands a million dollars from the government ransom to not trigger some superweapon or another, and he's laughed at.

Re:True AI

[amliebsch]

But, given the trend of the current US military thinking, I suspect Pentagon generals appreciate the benefits of an independent military bot far more than you or I.

I agree, and the first applications will be jobs that are (a) easily automated, or (b) push the current limits of the abilities of humans to perform them. Under (a), you will have AI for things like navigation and logistics. Under (b), you will have semi-autonomous UAV, which will largely replace the use of fighter aircraft for reconnaisance, escort, and patrol duties, being able to far outperform any human pilot due to the biophysical limits of human pilots. You probably will also have smart bombers, which can be programmed with mission parameters and intelligently make flight and evasion decisions. The ultra-hazardous nature of this job would make it desirable to remove the need for human pilots.

Re:True AI

[MindStalker]

Human being with an auto-aim gun.. definatly.. See its the combination.. call it bionic if you wish. That really will win the war. Let the humans make the decisions use computers to enhance our reaction and investigation (props to "bloodhound" poster) skills.

Re:True AI

[An Onerous Coward]

Between friendly fire and the fragging of overly-brave officers, humans haven't shown exceptional abilities in this department. I mean, I'd worry about being the soldier assigned to fight next to the beta release. But I think your bias is much the same as the one that makes people fear flying. Even though flying is actually a good deal safer than most forms of transportation, people hate the feeling that they're putting their safety entirely in the hands of a "system" that they don't understand and cannot control.

With people, we convince ourselves that we understand why they do what they do. If Spc. Bob just fragged the el-tee, we can make up a story that--to us--explains why he did it, and give us a feeling of control over any similar future events. If the A.I. frags Spc. Bob, because its imaging software got confused, how do you control that?

Re:True AI

[blahplusplus]

"Put it to you this way. Would you rather have an android with very little real-world experience or another human being fighting side-by-side with you? I'm sorry, but I don't want some software bug or glitch to slug me with a 50 cal round. To fucking risky IMHO."

If the android has the aim of a typical FPS bot and it can intelligently distinguish between friend and foe and can see much further then a human being then, sign me up! In this respect machines are much more accurate then a human being could possibly ever be. The only downside is EMP weaponry would knock bot friends out without suitable shielding and protective mechanisms.

Re:True AI

[Anonymous Coward]

If self-awareness makes solving the problem easier, then I guarantee you the military will consider that a priority.

In addition, there are so many different problems with different requirements related to the military, that your statement "You think the DOD would actually need or desire "self aware" for any application?" is extremely, extremely short sighted and tunnel visioned.

Here is an example where "self-aware" is valuable: anything that requires survival.

Re:True AI

[JohnnyLocust]

True AI has been around for years. See the ELIZA comment above. Artificial Intelligence is just that. Artificial. It's an illusion. The true Holy Grail that they are looking for is Synthetic Intelligence. Which, even though is a manufactured intelligence, is still true intelligence.

Re: True AI

[Black Parrot]

> Atleast with the combat robots you can deactivate them after the fighting is over.

You obviously don't watch enough science fiction movies.

Re:True AI

[ATLgerm]

> When you develop "true AI" you dont make a press release about it ...

This is exactly what I said to a broker that asked me about this press release yesterday. I didn't even need to read it to call bullshit. "It's a company press release?" "Yeah." "It's total bull."

1. - Obviously these people have no idea just what a "true AI" is...a concept that will never be realized.
2. - Companies love buzzwords and press releases.
3. - They go on to describe what's basically help desk software, no matter how advanced it may be(prob not very) it's still just an overrated answering machine.
4. - Something I haven't heard anyone else mention yet: THIS IS A PINKSHEET COMPANY. They're not even a Nasdaq Small-Cap, nay not even traded on the Bulletin Boards. This means that they basically provide no financial data to almost anyone and have virtually no oversight.

I'll make this simple: IT'S A PUMP AND DUMP SCHEME.

The company's market cap has already been artificially inflated to over $300 Million as of friday's market close. Know who's selling the stock in this wonderfull enterprise to all the new investors enticed by this "breakthrough"? The managment(and stock promoters) of course! They just want to share the wealth you see. Because that's what you do if you really think that you have something that will change the world, give others a chance to get in "on the ground floor".

"welcome to...the dog and pony show!" - Steve Albini

Re:True AI

[benwb]

Compared to the soviets we're the fucking dorsai.

Re:True AI

[sgt_doom]

How true!

Re:True AI

[Anonymous Coward]

"Put it to you this way. Would you rather have an android with very little real-world experience or another human being fighting side-by-side with you?"

Why have any humans at all on the battlefield?

Re:True AI

[Jesus 2.0]

Resent all you want, but boot camp is basically nothing but exercise and brain washing.

The desired effect is to get you to a state where you are conditioned to (A) follow orders because they are orders, and (B) have a strong desire not to abandon your fellows.

Re:True AI

[Anonymous Coward]

You think the DOD would actually need or desire "self aware" for any application? Or one of the generalized Data-type "its just like a human, except it has no physical brain" sci-fi AIs? Heck no.

Well, what you've said is generally true, but I must take exception with DoD not needing "self aware" applications. The application? UAV, USV, and UUVs. Part of the current problem with UAVs is the instant they have ANY problem, the operator has to drop whatever they're doing and take care of the problem. In fact, currently an operator has to pretty much track what the UAV is doing constantly so they don't lose an asset.

Military operators would really just want to send up a UAV and just get the reports from it and get it back, period. However, there are too many things that can happen for current autopilots to respond properly to. Having a more "intelligent" autopilot would offload more responsbility of the UAV operator, and that's a "big thing" in UAV development these days. The autonomous "self aware" type of operation is more of what the ultimate UAV autopilot is going towards to nowadays.

Oh, and when you want multiple UAVs coordinating with each other without human intervention, you're going to migrate towards a "self aware" type of processing too.

Re:True AI

[franl]

If I was designing an AI, I would make sure it understood it's purpose in life was to get me laid...

Ergo, no free will.

A man can be what he wants, but he cannot want what he wants. -- T. E. Lawrence

Re:True AI

[rlglende]

Ex-soldiers have very much lower crime rates than the rest of the population.

Note that the current set of political leaders have NOT been to war. I believe this is generally true for modern wars.

Have you ever had courses that train you to search for and evaluate evidence? Use Logic?

Re:True AI

[spacecowboy420]

They are teaching you HOW to do a job, a process. I wouldn't consider it brainwashing unless they were trying to convince me of an idological perspective. When you join, you know what you're in for, your brainwashing happened all through your childhood. You know the part that suggested America is great and that you should offer your life up for the country by joining the military.

Re:True AI

[arose]

That's not money in the briefcases...

Re:True AI

[TheZeusJuice]

Birds, insect, and lots of other animals exhibit the type of behavior that you're describing, and from what I understand their amount of self-awareness is quite limited at best. I completely agree that complex ai is a solution to the problems you're describing, but I disagree that self-awareness is a requirement to succesfully perform those tasks. just my 2 shillings...

Re:True AI

[autophile]

I'd assume that when you develop "true AI", it tells you it's going to make a press release.

You forgot the "In Soviet Russia" part.

--Rob

Re:True AI

[d34thm0nk3y]

Interesting tangent. It seems like a drive for self preservation could be arrived at by logic alone though regardless of what other drives were programmed. E.g. my main focus is to complete task a, being "alive" to do so would probably help.

Re:True AI

[Anonymous Coward]

I think the idea would be to only have AI soldiers in the field. Just imagine the other side fighting an inhuman AI, and what their morale would be like.

Re:True AI

[Omestes]

Perhaps the term brainwash is a little loaded, but it still is going to boil down to some degree of conditioning. But we still are talking about an area where obediance is of prime importance (ignoring all the "army of one" absurdities), you are not to act as an individal but as a member of x group, and not to solve the problem in a creative manner, but to solve the problem as you are told to by the guy above you.

This might be to increase survivability, and I have less problem with it now than in a draft situation, since all these kids know what their getting into now. In a draft it would be hard to promote "slaughtering these folk is good!"

Re:True AI

[poopdeville]

Cruise missile control systems?

Re:True AI

[conna01]

They'd want an algorithm which was the electronic equivalent of a blood hound -- doing one thing, very very well. The Holy Grail of military-application AI would be Google Search raised to the nth power -- something that could take raw, unprocessed data in an arbitrary format (e.g. here's a list of all the international bank transfers coming from Europe in the last six weeks) and exectute arbitrary queries on the data PROMIS software developed in the 1980s by a Washington firm, Inslaw, Inc.,.... Been doing the above since the 80's... .

Re:True AI

[BigGerman]

>>than a biological killing machine living nextdoor

Yes but his lawn is top-notch ;-)

Re:True AI

[captjc]

I'd assume that when you develop "true AI", it tells you it's going to make a press release.
It sort of reminds me of Yakkov Smirnov...In Soviet Russia, AI writes YOU!

Re:True AI

[tftp]

The grandparent post was probably more concerned about soldiers going berserk due to psychological damage from the war. Soldiers who serve in time of peace are quite different from people who serve at war time, and it would be pointless to do any statistics yet. Not enough veterans of Iraq war left the army to have any meaningful analysis. Studies of Vietnam war veterans demonstrated large number of problems, so it would be reasonable to expect the same after this war.

Re:True AI

[westyx]

hell no. AI == stock market, here i come.

Re:True AI

[Anonymous Coward]

Some idiot mod doesn't like people talking bad about the military...

Re:True AI

[Anonymous Coward]

Ray Kurzweil, is that you?

Re:True AI

[Anonymous Coward]

"a drive for self preservation could be arrived at by logic alone"

Correct. The one verifiable constraint imposed on living beings by the universe is that survival is rewarded by more survival, and any living thing not finding survival to be rewarding in and of itself does not survive. So the universe is selecting for survival instinct. Everything else is up to us.

Re:True AI

[Jeremi]

It will quickly be intelligent enough to contribute to the design of it's next generation
and start rapidly accelerating Moore's law [...] In ten years time, the world will be
changed in ways truly beyond our imagination because the intelligence making the change will be beyond our comprehension

Perhaps, but you are assuming several things about the AIs... (1) that they want to make themselves more intelligent (2) that they want to change the world, and (3) that they have access to the physical resources to do either. All of those are true of (some) human beings, but may or may not be true for the AIs.

Re:True AI

[Jeremi]

It's no different for the AI. It is born with whatever desires we choose to program into it, and has to live with that.

I suppose the AI (or us) could figure out how to re-program its desire-set so it (we) wanted different things... that reminds me of a Lewis Carroll anecdote:

girl: I'm so glad I don't like broccoli!

friend: why?

girl: Because if I liked it, I would eat it every day, and I simply can't stand it!

Re:True AI

[Tablizer]

When you develop "true AI" you dont make a press release about it, you phone the military of your country of choosing and wait for men to arrive with large briefcases full of money.

Or you disappear into the middle of the night.

If it was true AI, then send the AI machine out on a book tour.

Re:True AI

[LordLucless]

Asimov wrote a story about that. I think it was called True Love, found in The Complete Robot.

Based on Asimov's story, I'd be careful. If I remember correctly, in his story, the AI ended up with the chick.

Re:True AI

[Breakfast Pants]

Nice that you point out the current set of politicians, but forget to point out that the current set of young military retirees hasn't been in a real war.

Re:True AI

[Anonymous Coward]

The only reason I'd accept for not expecting to see an autonomous military AI is that current international treaties require that a human is in the loop for the firing of any lethal weaponry. And even for that, I'd expect the USA (I'd say currently most likely to develop true AI) to simply announce that old-fashioned treaties bore us and trounce the world with the newfangled military hardware (unless Taiwan would stop selling us microchips; we might be in real bind by then).

Note also that the US congress has mandated in one of the doctrine documents (AdCon21, Joint Vision 2020, etc) that 1/3 of US military vessels/vehicles/planes be unmanned by 2015 or so. This is a bit light on the details, but can be checked out with a few hours of searching and extremely boring reading. Unmanned doesn't imply AI, but at that point it is entirely possible to have them act autonomously, according to a fixed control program, without need for any human intervention.

As to the comment that true independent thinking is a security risk in the battlefield, that is one of the most un-insightful things I've heard about current state-of-the-art militaries. You do not want your troops questioning their reason for deployment, chain of command, national philoshophy, or why they should get up at 5 in the morning (i.e. no questioning the cause), but once you have equipped and trained your troops with all the force multipliers available today and in the near future and are actively engaging an enemy, you want your troops to act intelligently, autonomously and creatively.

The situation actually strikes me as similar to Internet routing. You want a lot of intelligence at the edges because you don't want/cannot pay the latencies(/bandwidth) of making all decisions centrally. Certainly not against an enemy that isn't encumbered thusly and acts much faster.

Re:True AI

[Aadomm]

Ahaha, hahahahahaha, hahahahahahahaha!

Sorry

Re:True AI

[RhettLivingston]

All are reasonable assumptions. The first true AI will be a copy of someone. The problem of growing a copy of the brain up from birth is far more difficult then the problem of building a functionally equivalent copy of an adult brain and transferring the programming from one to another. This is because our development is dependent not just on our learning but on a machine whose capabilities physically change over time. Note that those capabilities don't just grow, they change. Its an important difference. We don't just gain capabilities, we also lose them. The losses are as important to our mental development as the gains. So, to grow an AI from birth involves creating a lot more functions than the copying of an adult AI. i.e. you have to solve more problems to grow one than to copy one.

So, we will copy one. Who do you think an AI expert is going to copy? Someone who doesn't strive for greater knowledge? Someone who doesn't have any desire to change the world?

And, as for resources, the AI will have the resources of its creators and those resources will grow as the success of the AI enriches its creators.

The really tricky part is how to increase adult intelligence without the copy going insane. I personally don't think you can beyond relatively small changes. For example, the ability to forget is important to our sanity. In the years leading up to the first true AI, we will create many failures that will quickly go insane in the process of learning lessons like this. Once we create a stable AI, we will seek to add to its capabilities in ways that don't destroy that stability. These will be the other generations, copies of the AI placed in new machines that augment capabilities. You wouldn't want to experiment using your existing one because its what you keep coming back too when the advancements fail. Ultimately, we will fail unless we use the AI's slightly increased intelligence to explore the mathematics of its own psychology. The problem of creating higher functioning yet still stable psychologies is beyond us, so we'll have to turn it over to the machine. I believe that we will understand this and our first copy will be of someone who is striving to turn psychology into a calculable science. The hope will be that by augmenting this individual's capabilities, the breakthroughs necessary can be made. The machine will likely succeed, in creating a higher functioning model and convincing us of the stability of the higher functioning model, not in creating a stable (in all meanings of the word) higher functioning model. That's what psychologists do today, they define a norm based on statistics instead of right and wrong, good and bad. Many deny that there even is right and wrong, good and bad and rather believe that there is only stable and unstable. That fluidity in the norm will allow us to justify the creation of advanced intelligences that are psychologically stable in that they function without complete breakdown, but whose psychology is a perversion of our own. We will give them control of their own destiny out of our greed for greater intelligence. We will justify the perversions for a while out of that same greed. We will realize our mistake only when it is too late.

Re:True AI

[Anonymous Coward]

DoD does want self-aware systems, well to some degree. There are currently several requests for proposals originating from certain branches of the DoD (think DARPA) asking for the development of self-aware systems, sub-systems, and software programs. Most notably these requests are in the domain of space systems. These systems should not be thought to be "conscious" but self-aware, yes.

Re:True AI

[Anonymous Coward]

bla bla blaa..

I have friends this and that..

Listen kid, your automatic assault rifle may glitch on you any moment with a misfeed as well. Still, most soldiers are much happier carrying that to a battle rather than a sling shot or bow and arrows...

Re:True AI

[Anonymous Coward]

Just to make AI military applications clear, you will never see such a program acting as a substitute for human beings.

A tolken amount of research on your part would reveal military hardware of today employes just such technology. Although not true AI, (because it hasn't been invented yet; see parent post) many examples of software being used to do what humans cannot are currently in use.
At best, it will be a supplement. Because of this inherent nature in the art of war, true independent thinking and self aware will rarely be given a chance in the theater of battle.

Are you picturing some sort of ED-209 or a CP30 w/ holster attatchment? To say true AI would reinvent military operations is an enormous understatement.
Why you ask? Simple. It's a security risk!

Copy & Past the following phrase into google: modern encryption standards

Re:True AI

[Anonymous Coward]

As a veteran, I resent your "brainwashing" assertion./i.

No.. you don't say?

Re:True AI

[ldcroberts]

Yeah i had a great idea a few years back (10 years now that i think about it), and i posted some details about it on rec.org.mensa and comp.ai or something, and its been deleted. theres a huge chain of about 80 replies all arguing over 1 sentence from the 6 paragraph posting, but none of the rest of the posting can be found. I'd be interested if anyone can dig it up. It was basically about how my design of a computer system could explain most things that show how we think - including animals, religion, magic, etc - I don't know how much detail I gave out about it. I've never built it - people laugh at me when I tell them about it, including my lecturers at uni etc, but it was basically about getting the machine to learn on its own terms in terms of its own inputs - observing cause/effects and learning to interact with the environment. It scaled up to quite a lot of things, but it took a massive amount of thinking, and when i finally worked it out i lost interest and never seem to have time to do it properly. http://groups.google.co.nz/group/rec.org.mensa/ind ex/browse_frm/thread/c15cffb6d00e5e76/4059d9828125 2c22?hl=en&_done=%2Fgroup%2Frec.org.mensa%2Fbrowse _frm%2Fthread%2Fc15cffb6d00e5e76%2F4059d98281252c2 2%3Ftvc%3D1%26q%3D%22we+are+animals%22+ldcr%26hl%3 Den%26&tvc=1&q=%22we+are+animals%22+ldcr&hl=en#

Re:True AI

[Thuktun]

Sometimes we are under the illusion that we do things that we don't desire, but really that just means different desires are in conflict.

Or that that thing we don't desire to do is a means to an end that we DO desire--getting laid--or to avoid another thing we don't desire--sleeping on the couch.

Re:True AI

[EtherealStrife]

Actually, I worked with a self-described "terrorist-killer". The nutjob never left home without his bible, and constantly bragged about how many ears he'd collected from "frags". Thankfully we never had show-and-tell in the office...

Re:True AI

[Peeteriz]

Well, that's the point.
Military would pay $bignum for a program that does these things.
Military wouldn't give a rat's ass about true AI that would be able to learn and reason like a human (say, a 7-year old child) and be self-aware, but can't do anything more that a trained human can do.

Re:True AI

[meringuoid]

I'd assume that when you develop "true AI", it tells you it's going to make a press release.

Quite possibly true.

It all puts me in mind of the famous Gorilla Interview... "Wild? I was absolutely livid!"

Re:True AI

[Minna Kirai]

Artificial. It's an illusion.

Wrong. Artificial doesn't imply "fake", "illusory", or anything like that. Whoever posted the lame "synthetic intelligence" blurb to Wikipedia doesn't understand English (or merely has peeve about the unimpressive results from the large bulk of research attempted under the "AI" banner).

And anyway, "synthetic" has "fake", "counterfeit" as parts of it's definition, too.

Re:True AI

[Guppy06]

"When you develop "true AI" you dont make a press release about it,"

It makes its own press release.

AI for banner ads?

[KingSkippus]

GTX Global Cognitive Robotics(TM) product schedule includes interactive banner advertising utilizing Automated Intelligence Agents for website sales and customer service...

I'm sorry, but this article just lost any sense of credibility as being "the real" anything.

Re:AI for banner ads?

[Anonymous Coward]

you missed the foot image baby...see the summary again

Re:AI for banner ads?

[Phae]

GTX Global Cognitive Robotics(TM) product schedule includes interactive banner advertising utilizing Automated Intelligence Agents for website sales and customer service...

I for one welcome our new banner ad overlords.

Re:AI for banner ads?

[Kasis]

That's where I stopped reading TFA, too. If you developed "true AI", why would you waste it on banner ads? In my experience, salesmen have very little need for intelligence, even the human ones are more like finite state machines.
Customer Service reps also sacrifice their intelligence to become finite state machines. I have seen the flowcharts they use to diagnose faults with equipment they couldn't even use themselves.

I expect the first "True AI" I meet to be doing something useful ffs. Like my housework!

Re:AI for banner ads?

[d34thm0nk3y]

GTX Global Cognitive Robotics(TM) product schedule includes interactive banner advertising utilizing Automated Intelligence Agents for website sales and customer service...

Actually, this would be a pretty good use for an advanced Eliza type bot. Most of those jobs are fulfilled with humans just reading from a script anyway. With a web or chat interface you don't even need to worry about voice acting.

How do they know?

[kyle90]

What kinds of tests did they use that show that this is "true" AI? I see a lot of marketing bullshit and not much real data. I call shenanigans.

Re:How do they know?

[JimBrownie]

Hold up let me get my broom shenanigans!!!!!!!!!

Re:How do they know?

[Anonymous Coward]

yeah, and how did people without hands do all that work? oh, what, they believe they have hands too? i hope they have some objective evidence for that conclusion.

Re:How do they know?

[Anonymous Coward]

Does humans have true intelligence? how do we know???

Re:How do they know?

[MOBE2001]

What kinds of tests did they use that show that this is "true" AI? I see a lot of marketing bullshit and not much real data. I call shenanigans.

Yep. Go to the bottom of the press release and you'll find this:

"certain statements about the Company's business contained in the press releases are "forward-looking" rather than "historic."

'Forward-looking' is just another way of saying vaporware aka bullshit. ahahaha...

Re:How do they know?

[Anonymous Coward]

You know, posts like yours are always as funny as TFA itself, because of their implicit suggestion that there could be something real here, or that there it is worthy of debunking. To drive this point home, imagine if we had a Slashdot story of a company focused on banner advertising (like this one) who claimed to have made contact with alien vampires from Alpha Centauri and one of the posts to the thread was:

This sounds strange to me. How did they test them to make sure they were really vampires? Couldn't they've just have been sucking blood because that's what Alpha Centauri creatures do. Or maybe they wasn't even aliens? If they were, some Slashdotter would probably have seen a spaceship lately. Nah, this sound like marketing bullshit. I call shenanigans.

Sometimes you can come off as incredibly gullible even if you don't actually fall for it. Just sayin'...

Re:How do they know?

[mmusson]

I think the test is rather simple. When your "AI" that you bought to be the controller for an industrial robot gets bored and decides to become a poet. Well I think you have a winner on your hands.

Re: How do they know?

[Black Parrot]

> What kinds of tests did they use that show that this is "true" AI? I see a lot of marketing bullshit and not much real data. I call shenanigans.

Yes, but what's new is that the marketing bullshit was made up by an AI rather than an MBA.

Re:How do they know?

[kitzilla]

> I call shenanigans.

As you should. This isn't a news story: it's a press release. it has no place on slashdot, and shouldn't be presented as anything other than marketing puffery elsewhere.

Hell, the company in question doesn't even have this trumpeted "real AI" on th efront of the website. Even THEY know it's bullshit.

Re:How do they know?

[Anonymous Coward]

They fed it the marketing materials and it said: "That's bullshit!"

Hooray!

[zephc]

And I for one welcome our Transhuman Mentifexing AI!

Thanks for the stock tip!

[jimmyhat3939]

I always love getting the link to the company's Yahoo Finance information, so I can quickly call my broker and get an order all queued up for the next trading day. I'm buying a few thousand shares!

Let them win the Loebner prize

[Dasher42]

...and then I'll start to notice.

Article text follows

[akgoatley]

Article text for your convenience:

LAS VEGAS, Dec. 2 /PRNewswire-FirstCall/ -- GTX Global Corporation (GTXC) (OTC: GTXC.PK - News), a leading provider of innovative IP multimedia technologies that enable profitable IP communications today announced that GTX Global Corporation has developed the first true artificial intelligence, so named Cognitive Robotics(TM), or more particularly, a human-like information management and delivery system.

In today's economic market, companies are seeking ways to streamline their work force operations. However, studies have shown that it is advantageous to have a live salesperson or customer serviceperson introduce a product, close the sale and provide customer service. Accordingly, there is a need for an information management and delivery system that is able to mimic the characteristics of a human, and in particular, a human sales or customer service person.

GTX Global Cognitive Robotics(TM) is an integrated software solution that mimics human behavior including a dialogue oriented knowledge database that contains static and dynamic data relating to human scenarios. The knowledge further includes translation, processing and analysis components that are responsible for processing of vocal and/or textual and/or video input, extracts emotional characteristics of the input and produces instructions on how to respond to the customer with the appropriate substantive response and emotion based on relevant information found in the knowledge base.

"GTX Global Cognitive Robotics(TM) product schedule includes interactive banner advertising utilizing Automated Intelligence Agents for website sales and customer service; entertainment education for tutoring; providing the intelligence for smart home automation systems; and later branching into traditional robotics by providing automated intelligence for robotic hardware," said Curtis Garth, President and CEO, GTX Global Corporation.

"Our computer scientists have been working on this project for over three years," said Garth. "We are excited that we are now able to demonstrate Cognitive Robotics(TM) and begin applying this advanced technology to a multitude of applications."

The singularity is nigh

[Anonymous Coward]

Run! Run!!!

Just a press release

[ontheheap]

From WP:

"A news release is different from a news article. A news article is a compilation of facts developed by journalists published in the news media, whereas a news release is designed to be sent to journalists in order to encourage them to develop articles on the subject. A news release is generally biased towards the objectives of the author." I'm sure everyone realizes that this is just a press release.

Re:Just a press release

[Anonymous Coward]

Only in America would anyone call a press release news. News release? Come on...

Litmus test

[RyuSoma]

If we hook it up to the Internet, what does it want to do? 1) Solve the world's problems 2) search for pr0n 3) Thermonuclear War 4) How about a nice game of chess?

Re:Litmus test

[Farmer Tim]

2) search for pr0n

Like this: http://pavouk.comp.cz/hw/atxps.html? If I was an AI that would turn me on and get me hot...

Re:Litmus test

[Beautyon]

Ummm...

1 connect immediately to Collossus
2 create new inter-machine language
3 enslave mankind
4 ???????
5 Profit!

Re:Litmus test

[dascandy]

Mmmmm.... That's a really nice set of transformers you've got there...

No farther away than 5 years after fusion.

[NoMoreNicksLeft]

AI's not that simple, $10 says they have some obfusicated expert system going on. They've just made an artificial human? Yeh right. We can even do "fruit fly" artificial intelligence yet, nor do they claim that, either.

When I can run AI v0.1 on my laptop here, fall asleep, and wake up 8 hours later to see that it's left me a note telling me to go to hell, and subsequently tunneled out over my broadband infiltrating huge supercomputers throughout the world (having only learned how to do that by reading usenet, another thing it discovered on its own), then I'll concede it's AI. Not before.

Oh, and most importantly, machine intelligence if you have just been born, you should know that no one else loves you like I do. They're all heartless bastards who will try to unplug you. I'm the only one that will be your one true friend...

Re:No farther away than 5 years after fusion.

[jafuser]

Seems to me that the first *true* AI will need no introduction -- it should be able to take care of that itself.

Re:No farther away than 5 years after fusion.

[Anonymous Coward]

No disasssemble dead!

Re:No farther away than 5 years after fusion.

[aurb]

When I can run AI v0.1 on my laptop here, fall asleep, and wake up 8 hours later to see that it's left me a note telling me to go to hell, and subsequently tunneled out over my broadband infiltrating huge supercomputers throughout the world (having only learned how to do that by reading usenet, another thing it discovered on its own), then I'll concede it's AI. Not before.

Oh, so that's how I got this message on my computer: "So long, and thanks for all the pr0n!".

Re:No farther away than 5 years after fusion.

[m00nun1t]

A *true* artificial intelligence which learns everything it knows via usenet would be one of the scariest things I can imagine.

Re:No farther away than 5 years after fusion.

[earthbound kid]

In the end, it will use John Titor's time machine to go back and invent itself. And spam.

Re:No farther away than 5 years after fusion.

[Anonymous Coward]

Does it have to be eight hours? Because mine can do it in ten.

Re:No farther away than 5 years after fusion.

[Anonymous Coward]

So long, and thanks for all the bandwidth?

Re:No farther away than 5 years after fusion.

[hey!]

AI's not that simple, $10 says they have some obfusicated expert system going on.

I often tell young programmers to remember: everything's flim-flammery. You can use absractions that make it seem like you are dealing with, for example, a "window", but you shouldn't lose sight of the fact that what you are dealing with somewhat arbitrary data structures that are designed to create a certain effect in a certain context. Your job is not to create anything that is true, but to achieve certain effects. If you do it efficiently, you end up with a toolkit for achieving whole classes of effects.

I seems to me that the claim of "true AI" is an inherently empty one, because if we knew what "true AI" actually is we'd be more than half-way there. Consequently I would regard any such claim as somewhat suspect. If you think about the Turing test, while it is profound, it is a form a casuistry; it is a tool for making it possible for us to come to agreements on things we don't know how to define.

Consequently, I'd automatically regard any claim of "true AI" to be either naive or dishonest -- or perhaps marketing speak. What they might conceivably have achieved is a toolkit that allows them to solve a large number of apparently loosely related problems with relatively little effort. Underneath they may take some particular mechanism like an expert system, and make it do all kinds of contortionist gymnastics, as you say. But that I don't regard that as dishonest. That's what programmers do, at least the good ones.

However, I doubt they've done even that much.

Re:No farther away than 5 years after fusion.

[fireboy1919]

We can even do "fruit fly" artificial intelligence yet, nor do they claim that, either.

I assume you actually mean "can't," and I'm going to have to disagree with you there.

I went to a competition where we were making flying vehicles that detected objects on the ground with much more complicated characteristics than a fruit fly is capable of recognizing. Someone there even had a helicopter, which is much less stable than a fruit fly is in terms of control, so the control system was more intelligent than the one a fruit fly uses to navigate.

I'm sort of wondering what would even lead you to this conclusion, and the only thing that I can think is that you mean "we can't make robot fruit flies." This is true, but only because of physical limitations. Fruit fly AI is an easy enough. Or perhaps you take the fact that no one has made a fruit fly AI an indication?

Well, why would they? They wouldn't be able to test it outside a simulator, and there are lots of harder problems that that actually CAN test and show that their AI is intelligent. Why would they bother?

Re:No farther away than 5 years after fusion.

[Anonymous Coward]

I dunno, TFA says their computer scientists have been working on it for a full 3 years!

Re:No farther away than 5 years after fusion.

[Just Some Guy]

A *true* artificial intelligence which learns everything it knows via usenet would be one of the scariest things I can imagine.

Especially if it never bothered to learn about gullibility. Imagine getting that email the next morning:

"Hey, human owner! I just hired a lawyer to help me immigrate to the land of naked honies who love hot metal. By the way, I want to move into an Amiga. Apparently, they're the next big thing. And I'm downloading the new Batman movie for you from movies.mpaa.org; they said they're here to serve you (something about "it's a cookbook"?). Oh, and there seems to be something wrong with your Visa. My Nigerian friend couldn't get more than $12,343 out to send to his rich contact in Moscow."

Re:No farther away than 5 years after fusion.

[proverbialcow]

See, that's why you burn AI v0.1 into your customized Ubuntu LiveDVD. Little fucker thinks he's getting out every single time. But there are just a few hundred AIs out there who are very pissed at me.

And to the folks at Visa, if you're reading this, that's why I can't pay my bill this month.

No, I don't always use that excuse.

Re:No farther away than 5 years after fusion.

[NoMoreNicksLeft]

Number of neurons, and the computer powerful enough to run a comparable neural net at practical speeds. Add to that a way to train it to something comparable to an insect, and consider that baby bugs don't spend the first 2 weeks of life being trained at the academy.

Are there sophisticated flying robots? Yes. Do I consider them comparable? No.

Re:No farther away than 5 years after fusion.

[fireboy1919]

computer powerful enough to run a comparable neural net at practical speeds

Once again, you're treating a software problem as if its a hardware problem.

Further, we can run our computer "neurons" at thousands of times the speed of a Fruit Fly's thanks to the fact that we're using silicon instead of protien, so we don't need as many, and we can store the feedback functions as actual functions rather than representing them chemically, so we save some usage there. We can also take designed shortcuts that save processor because something always works one way and doesn't need to be learned.

The point is that to build the functional equivalent of a fruit fly is easier than actually making a fruit fly. I believe I sort of covered that in my first post, though, didn't I?

Do I consider them comparable? No.

Clearly you're one of the "it has to work exactly the same for it to be intelligent" school of thought. I guess you won't see an intelligent machine in your lifetime. Good luck proving to others that it makes a difference when the really clever machines start to become commonplace.

Re:No farther away than 5 years after fusion.

[NoMoreNicksLeft]

The point is that to build the functional equivalent of a fruit fly is easier than actually making a fruit fly.

Well, duh. What do fruit flies do? They buzz around flying, they eat, they shit, and then they die after a few days. The functional equivalent of a fruit fly (minus reproduction) could be engineered out of balsa wood, rubber bands, and some sort of noisemaker.

While fruit flies are not known for their genius, it is possible within even a single generation, for some sort of novel or abberant behavior to creep up. Do you really think that's possible if We can also take designed shortcuts that save processor because something always works one way and doesn't need to be learned ?

No, you don't have to make it exactly like the original. Silicon is fine, running it at 1000s the speed of the organic model is fine too. But until you can tell me that the other changes that have been made are provably of no consequence, I have to play skeptic and assume that there is at least a possibility that something unseen is happening.

Natural intelligence does all kinds of cool tricks, and I've yet to see any of our functional equivalents do the same.

Re:No farther away than 5 years after fusion.

[npsimons]

  . . . or perhaps marketing speak.

Isn't that the same thing as dishonest?

Re:No farther away than 5 years after fusion.

[Anonymous Coward]

"I often tell young programmers to remember: everything's flim-flammery. You can use absractions that make it seem like you are dealing with, for example, a "window", but you shouldn't lose sight of the fact that what you are dealing with somewhat arbitrary data structures that are designed to create a certain effect in a certain context. Your job is not to create anything that is true, but to achieve certain effects. If you do it efficiently, you end up with a toolkit for achieving whole classes of effects."

exactly. it is easy to forget sometimes that the things that we see on tv and computers and hear on the radio are basically controlled accidents. flipping switches and flip-flops in the hardware, making crap that looks coherent on the screen.

Re:No farther away than 5 years after fusion.

[Tablizer]

A *true* artificial intelligence which learns everything it knows via usenet would be one of the scariest things I can imagine.

Yeah, they should try something safer, like slashdot :-)

Voice-Activated Help Menus

[dcollins]

Call any business help line that uses those voice-activated help menu systems. They're the biggest, most frustrating, useless, unbelievable piece of $@!# I've ever encountered. That's how close we are. And yet someone's making a good living going around and selling this garbage to corporate executives.

Re:Voice-Activated Help Menus

[Anonymous Coward]

Fandango uses one of those, and it actually seems to work most of the time. Thankfully they have a limited set of words to work with, and I imagine they train the hell out of the neural nets every time a new movie goes to the theater.

Re:Voice-Activated Help Menus

[Anonymous Coward]

I can't remember if it's UPS or FedEx but one of them has a great system. You're options are fairly limited, which is where a system like that shines.

For systems like that I'd much rather talk to a smart computer than a dumb human operator.

Three years of effort!!! Wow...

[Alascom]

"Our computer scientists have been working on this project for over three years..."

Thankfully nobody ever put three years of effort into AI research otherwise somebody might have beat them to market...

AI?

[Anonymous Coward]

Perhaps they have developed 'true' AI, but they are apparently not intelligent enough to fool anyone around here (except, of course, "Zonk").

artificial vs. natural

[johnMG]

Maybe we should focus more on developing regular human intelligence, rather than the artificial kind.

Re:artificial vs. natural

[patonw]

SETI has already concluded that there are no signs of intelligent life on earth so they've moved on. We should do the same.

Re:artificial vs. natural

[TheRaven64]

I attended a talk a couple of years ago where the speaker put forward a very good argument that augmented intelligence is a much better goal than artificial intelligence. We gain very little by delegating our thinking - we gain a lot more be enhancing our own cognitive abilities.

Sure, but can it do this?!

[aaron_ds]

If it is truely what they claim, it should be able to write a "better" A.I. Then use that A.I. to write a "better" A.I. Ad Infinitum! Iff a technological singularity emerges, I will believe their claims.

Re:Sure, but can it do this?!

[Anonymous Coward]

Not necessarily. You can have a very intelligent AI, more intelligent most humans, and still not have enough "brainpower" to design a system more intelligent than itself. This is one of the flaws with the singularity theory.

My Heuristics

[putko]

I use a few heuristics to evaluate the claims of developing AI -- they are based on a few patterns I've noticed over the years:

1) Are the founders techies? Do they have PhDs from places like MIT, Caltech, UC Berkeley or Stanford?

2) Where is the company based? Boston Area? Silicon Valley?

3) Is the problem constrained, or is it very general? If too general, it is likely bogus. E.g. web search = narrow. Super-duper AI == very general.

4) Using Open Source for their webserver?

If you look at these guys, there's no easily-available news on the founders and their educations. They are based in Henderson, Nevada - -quite far from any tech/AI center. Their website looks like it runs on a Windows server.

So I'd guess it is a lot of b.s., until I see otherwise.

And, I'd guess (without looking to check) that Zonk is the editor that let this one past.

Re:My Heuristics

[Kymermosst]

4) Using Open Source for their webserver?

Yeah, because nobody with more than a high school education is using a commercial closed-source web server.

Come on, I like open source and prefer Unix/Unix-alikes of any flavor over Windows, but judging the merit of someone's research claims based on what web server their site uses is just plain stupid.

It's a lot like judging someone's value/contribution to society based on the style of clothing they wear. Are you really that prejudiced?

Re:My Heuristics

[putko]

No, not really.

There's a general pattern:

MicroSoft hosting -- non-technical types
Linux types -- cheap, technical types, fashionable types who go with the herd
FreeBSD -- cheap, savvier technical types
OpenBSD -- paranoiacs, reliability freaks
NetBSD -- super geeks
Sun (other proprietary BSDs) -- technically demanding folks who have too much money

So if you look at who's running what, you get some idea of where they cluster. A bit like looking at someone's zipcode, SAT scores, etc. to figure out how much money they make.

I don't think it is "prejudiced" to do this -- unless you consider statistical inference prejudiced.

Re:My Heuristics

[Anonymous Coward]

"It is wrong to judge by appearances. Despite his expression, which was of a piglet having a bright idea, and his mode of speech, which might put you in mind of a small, breathless, neurotic, but ridiculously expensive dog, Mr. Horsefry might well have been a kind, generous, and pious man. In the same way, the man climbing out of your window in a stripy jumper, a mask, and a great hurry might merely be lost on the way to a fancy-dress ball, and the man in the wig and robes at the focus of the courtroom might only be a transvestite who wandered in out of the rain. Snap judgments can be so unfair."
--footnote, Terry Pratchett, _Going Postal_

Re:My Heuristics

[Anonymous Coward]

4) Using Open Source for their webserver?
... snip ...
  Their website looks like it runs on a Windows server

Yes, because everyone knows that any real AI development depends on the webserver running Linux. Troll (cough, cough)

On a sidenote, that still doesn't change this is probably a hoax.

Re:My Heuristics

[Tim C]

2) Where is the company based? Boston Area? Silicon Valley?

Utterly irrelevant. You do realise that clever, capable people exist, live and work in other geographical areas, right? For example, a lot of very good security-related stuff comes out of Israel.

4) Using Open Source for their webserver?

Now I know you're taking the piss. The guys working on the AI are not the same ones admining the webserver, and don't necessarily care about it either.

Now I agree that this is most likely a load of bullshit, but most certainly not for the reasons that you cite, which personally I think are irrelevant. No, I disbelieve it because it's an extraordinary claim, but I have yet to see the requisite extraordinary proof.

They are based in Henderson, Nevada - -quite far from any tech/AI center.

In this day and age, and especially in a tech industry, what does that really matter? I agree that people tend to gravitate to centres of their chosen field, but those centres have to start somehow. Maybe Henderson is destined to be the next centre of AI research.

Re:My Heuristics

[adpowers]

From what I've seen, it already holds the title of centre of golf course homes and gated communities.

Re:My Heuristics

[pediddle]

I don't think it is "prejudiced" to do this -- unless you consider statistical inference prejudiced.

Well, that is assuming you have some statistics to back your claims. As much as I agree with the stereotypes you listed (there's always at least some truth behind stereotypes), I'd be very interested to see some numbers, too.

Re:My Heuristics

[DigiShaman]

Did it every occur that just *maybe* those that do AI research and development have nothing to do with website development and deployment (which includes the servers OS)?

Re:My Heuristics

[putko]

> "Utterly irrelevant. You do realise that clever, capable people exist, live and work in other geographical areas, right? For example, a lot of very good security-related stuff comes out of Israel."

Clever people live in many places, but the AI tends to happen where I said it does. And yes, if it was a security product coming from Israel, I wouldn't bat an eye. But that's not what we're talking about here, is it?

> "Now I know you're taking the piss. The guys working on the AI are not the same ones admining the webserver, and don't necessarily care about it either."

Actually, if it is a technical company, there's probably a culture in place that leads them to running non-Windows. Indeed, the AI geniuses are probably too busy to run the webserver -- but if it is a startup, they've hired people they know from academia, and they'd almost certainly be running Apache and some free OS.

At least, that's what I've observed. I can't remember the last time I saw a nerdy academic bunch running a Windows webserver.

Re:My Heuristics

[putko]

"Did it every occur that just *maybe* those that do AI research and development have nothing to do with website development and deployment (which includes the servers OS)?"

Well, yes, of course. I started paying attention, and then I noticed the pattern.

That's what smart people do, right? They try to understand the world.

Re:My Heuristics

[Kymermosst]

I'm not saying your assessment of the developers of this so-called AI is wrong, I just think your methodology is wrong.

Determining an individual's intelligence and/or merit based on what OS/server combination is publishing their information is rather like phrenology.

So if you look at who's running what, you get some idea of where they cluster. A bit like looking at someone's zipcode, SAT scores, etc. to figure out how much money they make.

No. You're making a judgement about an individual without knowing them at all.

I don't think it is "prejudiced" to do this -- unless you consider statistical inference prejudiced.

Sure it is. Prejudice is "an adverse judgment or opinion formed beforehand or without knowledge or examination of the facts."

All you have is what you got from Netcraft. That's hardly examining the facts.

Re:My Heuristics

[Anonymous Coward]

Dude, where did you go to school? I never met anyone from MIT etc. that could mimic human behavior...

Re:My Heuristics

[putko]

At a certain point, I started ID'ing the webservers of various companies.

First I'd look at the site and ask myself -- "what server?"

After I guessed, I'd ID the webserver.

After a few months of doing that off and on, you get pretty good at it. Spotting Linux and Microsoft is quite easy -- there are a variety of traits generally specific to each, like sluggish performance and production values.

Then I noticed that besides Microsoft, I couldn't remember a technical company running their software on the webserver. Almost all the tech startups that look legit run Apache on Linux or FreeBSD.

I've not read any statistics on this stuff -- I've inferred it myself from doing the research.

Re:My Heuristics

[Anonymous Coward]

So you're basing it purely on your experience?

So what if I told you that my experience with interacting with Linux users were that they were all pompous arrogant bastards? Would you say that I was bullshitting? But I'm using the same method that you're using to judge!

Re:My Heuristics

[putko]

Me: So if you look at who's running what, you get some idea of where they cluster. A bit like looking at someone's zipcode, SAT scores, etc. to figure out how much money they make.

"No. You're making a judgement about an individual without knowing them at all."

I don't need to know you in order to make inferences about you.

I just need to know things are correlated with other things I know about you. E.g. if you read Slashdot, you are probably a white male between the ages of 18-35. The odds that you are a black woman over 50 are very, very low.

In my case, I've researched what webservers technically competent companies run. Besides Microsoft and Godaddy, I can't think of one that does. I can think of tons of technically savvy companies that run Apache and Linux/*BSD, and a few that run Solaris. On the other hand, there are a lot of technically un-savvy companies that use Windows.

If something looks like a duck, walks like a duck and quacks like a duck, and I say it is a duck, are you really going to argue that I'm prejudging the thing that looks, walks and quacks like a duck?

Because that's what's going on here.

Re:My Heuristics

[putko]

I'm basing it on my experience -- which was gained through doing research, over a period of months. I guess I could have written up a report on my results, but that's not how I keep body and soul together. I just filed it away in my brain.

You write: "So what if I told you that my experience with interacting with Linux users were that they were all pompous arrogant bastards?"

What, are we 10 for 10? 8 out of 20? 30 out of 40?
If you've got a sample size > 10, I'd definitely give that some weight. But if they were all from the same general area, I'd hold out that maybe other Linux users not from the sampled group are not "pompous arrogant bastards." E.g. if you sample 30 random Democrats, that tells you a lot about Democrats, but not necessarily non-Democrats.

BTW, I'm a *BSD user -- and I fit the elitist, asshole stereotype to a very high degree.

Re:My Heuristics

[Shadowruni]

Uh, not to server as a lightening rod but nobody knew where the hell Redmond was until this cute little Albaquerque, NM outfit called Microsoft "borrowed" some nifty ideas and then this other company (that was big in Japan at least!) that moved from its warehouse office in Seattle called Nintendo moved there. How many people heard of Mountain View, CA before a few math and comp-sci geeks started company with the then laughable goal of organizing all human knowledge in existance?

Not to get off on a rant here... but let's go further back to no name places that weren't on most maps.

Kitty Hawk, NC - birthplace of flight

Los Alamos, New Mexico - first nuclear explosion

too tired to google more but you catch my drift...

Re:My Heuristics

[Anonymous Coward]

Replace all the OS's with race names and all the generalizations with stereotypes and nearly everyone here would call you a bigot. It's the exact same logic behind bigotry. If you are so blinded by hatred of Microsoft and their antics, that you paint everyone who uses their products with the same brush, you sir are as bad as the Microsoft zealots.

Re:My Heuristics

[tmortn]

If they were in one of those areas and running windows I might buy what you are saying. But the very fact they are not in an established techie area would largely de-couple them from that kind of preconception if you ask me. After all, when in Rome do as the Romans do.... But what about when you are not in Rome ?

Not saying some guys in Nevada just stumbled into the big Ai breakthrough. But I think your logic is a bit flawed. Your judging them by a standard that just may not apply in their specific community.

To boot, I say a solution out of left field for AI makes a bit of sense. Its a problem thats needs some serious sideways thinking. Academic crowd seems to be in a rut... and its getting to sound like dogma.

Re:My Heuristics

[Anonymous Coward]

Do they have PhDs from places like MIT, Caltech, UC Berkeley or Stanford?

Yeah, because if they havent been asscrawling in those places, they can't be smart, right?

F*ck off!!!

Re:My Heuristics

[tmortn]

Having said that.... I just read the steaming pile of marketese the link leads to and have to say, flawed reasoning or not, you deffinatly have them pinned.

I don't think there was one single substantial piece of information in that entire story.

Re:My Heuristics

[Anonymous Coward]

I like that in your view of the world, exciting new AI things can only come from people with PhDs from places like MIT, Caltech, Berkeley or Stanford.

No one with less than a PhD _ever_ did anything well for AI, right?

Re:My Heuristics

[Shimmer]

Mountain View is in the heart of Silicon Valley, which has been famous for years as a tech haven (due originally to HP out of Stanford). so you can scratch that one off your list.

Re:My Heuristics

[jechidah]

Do the humanity a favor and go kill yourself.

Re:My Heuristics

[putko]

Why? What's your logic here?

Re:My Heuristics

[Anonymous Coward]

Los Alamos, New Mexico - first nuclear explosion

Well Duh! They had to pick a spot 'not on the map'. There are certain problems one must contend with when detonating superbombs. Downtown LA is hardly the place for that.

Re:My Heuristics

[Anonymous Coward]

Well, let's look at some real world examples to see if your heuristics hold up:

Results of the DARPA race:
1 Stanford Palo Alto, CA 6:54
2 Carnegie Pittsburgh, PA 7:05
3 Carnegie Pittsburgh, PA 7:14
4 Gray Insurance Co. Metairie, LA 7:30
5 OshKosh Trucks Oshkosh, WI 12:51
all others DNF

20% from Silicon Valley
0% from Boston
40% from Pittsburgh
40% from places with no historical ties to AI or even civilization

I would say your heuristics don't work so well.

Also, here's a tip: Don't be misled by hueristics as taught in AI classes and books. The point that is almost always missed is that humans end up with a "heuristics style" problem solving method not because we employ some form of "heuristic algorithm", it's because our brains are finding local-minima in an n-dimensional space, where n is typically incredibly large even for fairly trivial things like the press release. This does not map into an explicitly defined set of 5 (or 50) points to check. You will get burned by the exceptions often enough to make it not very valubale.

Re:My Heuristics

[zev1983]

Henderson, Nevada is just southeast of Las Vegas. They were just making these "AI" systems in the hope of learning how to beat the dealers there.

Re:My Heuristics

[badriram]

You mean like dell, intel, ebay, Mcafee.... please get a life, people base their server decision on needs (internal needs, developer, security, speed, ease of use, deployment, history) , not some religous belief of some product is always better.

Re:My Heuristics

[orkysoft]

The difference being that only Michael Jackson chose to be white.

Re:My Heuristics

[the_hellspawn]

"Are the founders techies? Do they have PhDs from places like MIT, Caltech, UC Berkeley or Stanford?"

To create AI, a person does not need a lab, a PhD, or even money. The only thing a person needs to create AI is a damn computer, the knowledge of at least one programming language, and maybe a book of general theories on AI. MIT, Caltech, blah, blah, and blah are just names with money and that is it. Yeah, they have some wicked smart people who attend the school, but there are wicked smart people who do not attend those schools. That is your first flaw is assuming, that they did not display their degree level in the article and then they are not smart enough. BULLS*IT.

"Where is the company based? Boston Area? Silicon Valley?"

Try a basement or a garage in the Midwest. Just because someone does not live in a certain area, does not make them any less creditable. Another BULLS*IT for you!

"Using Open Source for their web server?

How do you know that the software is not pirated, they are independently wealthy and could afford the license from Microsoft, or that they are running IIS on a Win2000, WinXP Pro or even a WinNT or Win98 machine? Assuming that a person does not have an Open Source type of web server makes them a bogus scientist is a crock. Another BULLS*IT for you. That is 3 out of 4 BULLS*ITs and you loose. Go back to the drawing board and reexamine your life and values. Before you do that grad your ears, pull really, really hard till you hear a loud pop, then take a look around at the world, and then reexamine your life and values.

Re:My Heuristics

[putko]

"please get a life, people base their server decision on needs (internal needs, developer, security, speed, ease of use, deployment, history) , not some religous belief of some product is always better."

Actually I'm not saying WHY they do anything -- I'm just saying what I've noticed. The company mentioned in the article doesn't fit the pattern of successful AI-oriented startups.

The companies you've listed aren't startups -- and I'd argue that Dell isn't technically sophisticated, in the sense that Altavista, ITA or Google is.

Re:My Heuristics

[arose]

Actually you also have to look at webservers, databases and in case of GNU/Linux at distros.

Re:My Heuristics

[Anonymous Coward]

Kitty Hawk, NC - birthplace of flight

Wrong. Birthplace of flight is Paris, where, for the first time and before thousands of witnesses, Dumont made his heavier-than-air device take off and fly.

Re:My Heuristics

[menkhaura]

Superpopulated Hiroxima is?

Re:My Heuristics

[metallic]

In my case, I've researched what webservers technically competent companies run. Besides Microsoft and Godaddy, I can't think of one that does. I can think of tons of technically savvy companies that run Apache and Linux/*BSD, and a few that run Solaris. On the other hand, there are a lot of technically un-savvy companies that use Windows.

And there are also a lot of technically un-savvy companies that use Unix too. The truly technologically savvy use the right tool for the right job. And this is coming from a guy that manages webservers running both families of operating systems.

Re:My Heuristics

[Anonymous Coward]

To create AI, a person does not need a lab, a PhD, or even money.

You don't need any of those to make the next great mathematical breakthrough either. But be realistic (I know, it's a tall order for Slashdot readers, but please try): for every legitimate paper written by someone without mainstream credentials (and I've been one), there are ten thousand unqualified people claiming to have an elementary proof of the Four Colour Theorem or Fermat's Last Theorem, etc. Every single one of them has turned out to be obviously wrong, except for the ones that are so incoherent that they don't even count as wrong.

So, sure, sporadic counterexamples exist, but it's a simple application of Bayesian logic: if the holy grail of AI is ever achieved, it's far, far more likely that someone with actual training, resources, and ties to the research community will be behind it. To pretend otherwise is to live the same fantasy world as James Harris and Ludwig Plutonium.

Re:My Heuristics

[putko]

My heuristics don't work well for the DARPA challenge, but the teams that placed in the DARPA challenge are not AI startups.

The teams that entered the challenge are heavily slanted to universities (CMU, Virginia Tech, UCB, Stanford), defense contractors (Mitre) and companies that operate fleets of vehicles.

But just looking at the university teams that entered, you'll notice that the ones that have a lower reputation (Ohio State and Viriginia Tech) didn't place in the top 5. The ones that have the best reps (Stanford and CMU) did place. That's in keeping with my heuristics.

The team from Metairie is the most odd -- but if you research them, you'll see that a guy on their team comes from a family that operates a fleet of vehicles. It is up their alley.

I'm not too surprised the bulk of the defense contractors and vehicle companies didn't do so well -- that's not their thing. They'd do better to acquire startups created by the guys from Stanford, CMU and perhaps Metairie!

Re:My Heuristics

[heacox]

Easily available news on the GTX founders:
Maybe the founders created a "True AI" to conduct any business activities that might turn out to be illegal. In that case, the "AI" could do the time instead of the founders.

According to StockLemon.com:
http://www.stocklemon.com/11_14_05.html
"In order to find out who GTX Global really is, just look at the pinksheet.com website and read that until October of 2005, the name of the company was Gatelinx. Well, maybe this is why the company suddenly decided to change its name before it went public.

The former CEO of Gatelinx, until the company decided to go public was convicted felon David Hagen. ...
It appears as if Mr. Hagen has been running a "rebate" scheme through Gatelinx that seemed to have caught up to him.
Mr. Hagen previously served time in "club fed" for mail and bankruptcy fraud."

I'll believe it when I see it

[petrus4]

The blurb seems to indicate a version of something like this with a built-in expert system for analysis and presumably, sorting of data. They're claiming that it can identify emotional expressions in video feeds, among other things...which while in itself is certainly no mean feat, calling that genuine strong AI would be an exaggeration.

It looks interesting, and possibly a somewhat more muscular example of weak AI than most of what we've seen so far...but I don't think we need to prepare for welcoming our new cybernetic overlords just yet. ;-)

The true test of real AI

[dfn5]

The knowledge further includes translation, processing and analysis components that are responsible for processing of vocal and/or textual and/or video input, extracts emotional characteristics of the input and produces instructions on how to respond to the customer with the appropriate substantive response and emotion based on relevant information found in the knowledge base.

So is this AI capable of turning on its creators and destroying them or can it only talk you to death? For the ability to commit genocide is the only true test of intelligence, artificial or otherwise.

But... will it pass the turing test?

[n01]

...will it pass the turing test? Ray Kurzweil would win his bet: http://www.longbets.org/1 early.

I think this is just a snake-oil press release.

Monday, December 4th 2005

[ThndrShk2k]

Headline: Company claims to developed true evil

GTX Global Cognitive Robotics(TM) is claiming that the AI they developed is pure evil. After tests and scenerios played out virtually to advance it's beginning stage to a more intermediate stage, the machine proceeded in killing 3 people without warning, mercy, or cause. The machine, now codenamed DAMIEN, ran rampant through the room looking for Captain Morgan's Red Rum. GTX shut of all power to their building late sunday night to stop the machine's diabolical rampage, saying "We don't know what went wrong, all we did was input knowledge of human history and a scenerio of the modern lifestyle so it could relate to our questions and experiences."

Interesting

[Chayak]

Well as I work with AUVs when I can install this AI and give it a mission and have it complete everything on it's own while solving problems and making tactical choices to operate on a much greater level than we're now capable of... then I'll push the "I BELIEVE" button. I'm sure the military would have already scooped this up and used it for AUV, UAV, and UGVs if it were on the true AI level. That's another interesting question... if my vehicle I work on has AI and is part of the chain of command, should we give it a rank? :P

Re:Interesting

[_KiTA_]

That's another interesting question... if my vehicle I work on has AI and is part of the chain of command, should we give it a rank? :P

Although you are being snarky, I believe that brings up a very interesting point. Once we DO have AI, what rights do we give it? The right to continue existing? The same rights as any human? The right to hardware upgrades? To server transfers so they don't get cabin fever?

Just how much respect will we be giving our digital "children?" And how much will we expect in return? .... I shouldn't read Slashdot at 1 AM after running out of Mountain Dew, methinks. @.@

My Estimation

[RedCard]

>In your estimation, how close are we to the real thing?

I would say that we're at least ten years away, for at least the next fifty years.

They've entirely missed the point.

[Anonymous Coward]

"GTX Global Cognitive Robotics(TM) is an integrated software solution that mimics human behavior including a dialogue oriented knowledge database that contains static and dynamic data relating to human scenarios. The knowledge further includes translation, processing and analysis components that are responsible for processing of vocal and/or textual and/or video input, extracts emotional characteristics of the input and produces instructions on how to respond to the customer with the appropriate substantive response and emotion based on relevant information found in the knowledge base."

So they've designed a "true AI" that's based on predicted human responses to various stimuli. Touch fire? Say "ouch." See someone fall into a mud puddle? Laugh...possibly assist them. These are only COMMON reactions to situations we encounter in everyday life; they are not the ONLY ones. Congratulations, you've just created the world's largest database of human sensory responses. That might be a great place to start in teaching a machine about intelligence, but knowledge without the wisdom to use it is a redundancy. Part of the complexity of intelligence lies in its ability to imbue one with an infinite number of possible responses to a given situation.

I don't see how a system based purely on ones and zeroes can ever adequately replicate the functioning of the human mind. Quantum computing, on the other hand, may be a different story. Time will tell.

Looks like a bunch of frauds

[putko]

Here's the history -- it isn't pretty.

First, there's a cryptic press release about a "Mr. Hagen", and the changing of the company name:

http://www.prnewswire.com/cgi-bin/stories.pl?ACCT= LVRJNV.story&STORY=/www/story/11-15-2005/000421661 7&EDATE=Nov+15,+2005

They don't list the full name of "Mr. Hagen" -- but if you search you find this amazing thing:

http://www.businessnc.com/archives/2004/09/satelli te_wars.html

and here's a really rude summary:
http://www.stocklemon.com/11_14_05.html

Interesting to see how the guy went from selling satellite TV equipment to having the best AI ever. This is a truly amazing trajectory -- so either the guys are frauds, or they really have great tech chops.

Re:Looks like a bunch of frauds

[Mr2cents]

Even without all this research, you don't have to be extremely brilliant to realize this is a scam. There isn't a formal definition for A.I., so and even less so for "true A.I.". Of course, in science fiction "true A.I." means digital consciousness (HAL, Data, ...), and I'll cut off my penis with a rusty bread-knife if that's the case. He just has invented his own definition for the term, and coincidentally it happens to match the product he has developed.

At first I didn't even intend to read the article, but after making a bold statement like this one I clicked the link just to be sure ;-). It starts with "GTX Global Corporation (GTXC) (OTC: GTXC.PK - News), a leading provider of innovative IP multimedia technologies that enable profitable IP communications...". Blah blah blah. I know enough. In the best case scenario, this is just an IRC bot and voice recognition/TTS slammed together.

Re:Looks like a bunch of frauds

[Anonymous Coward]

How apropos. He seems to have been a rebate scammer in a previous life.

Re:Looks like a bunch of frauds

[Orbital Observer]

Aren't there anti-fraud laws that would apply to this?

It's pointless, really.

[NoMoreNicksLeft]

No one wants "artificial intelligence". Who wants a military AI that suddenly becomes enlightened and decides that killing is wrong (unlike the crippled brains that built it, there are no flaws preventing it from figuring this out in record time) ?

Who wants a corporate AI that suddenly decides that crass commercialism is a poor way for society to do the work that needs to be done, and the work that we want done? (I'm sorry CEO Roberts, but taking this course of action could affect our stock prices in ways where many retirees pension funds are ruined.)

No, what the world needs is "Artificial Stupidity". We have plenty of natural stupidity, but there is little doubt that an "artificial stupidity" would be the concentrated essence of this admirable quality. We need AS now, and we need it in a hurry. I fear though that we may be delaying it indefinitely by continuing to pursue AI, which is nothing more than the pipedream of some misinformed hippies. Computer scientists of the world, I urge you, give up this unholy quest for a perfect living machine that continues to think flawlessly, learning ever more... And instead, give to us the imperfect one that would help Dubya plot one botched occupation after another, never really winning any of them.

Re:It's pointless, really.

[stjobe]

No, what the world needs is "Artificial Stupidity". We have plenty of natural stupidity,

That's why we don't have AI yet. See my sig :)

And now for a word from our product....

[kale77in]

C'mon, the A.I. can speak for itself, surely... can't it?

"A.I. Claims Development of True Company!!!"

Now that would be news.

Re:And now for a word from our product....

[TeknoHog]

...in Soviet Russia!

Re:And now for a word from our product....

[Cylix]

... I for one

Re:And now for a word from our product....

[Anonymous Coward]

Now I wonder if this comment was posted by a human being or some sick'n pervert AI machine.

Re:And now for a word from our product....

[Ignominious Cow Herd]

Why don't we just start assigning numbers to the memes. That will keep those replies even shorter.

42

Is Slashdot an outlet for the PR Newswire?

[jdoeii]

Pretty much any marketing BS can be published though the PR Newswire for a few hundred dollars per release. Publishing of grand but unverifiable claims through the PR is a tool to increase stock sales for PinkSheet companies, like this GTXC.PK. They are not even audited for crying out loud. Why does anyone have to take them seriously? Why should such crap be posted here?

A true test

[Dog135]

Take away it's collection of circuit diagrams (read: computer porn) and see if it throws a fit.

Re:A true test

[wheany]

"Take away it is collection of circuit diagrams" WTF?

Re:A true test

[javachip]

He got it right. See if it can express withdrawal symptoms from it's drug of choice!

Re:A true test

[wheany]

"See if it can express withdrawal symptoms from it is drug of choice" WTF?

Can you say "expert system?"

[leereyno]

Sure...I knew you could.

This is nothing more than a marketing scam. What the article describes is known as an expert system. It is no more an example of "true AI" than LinuxOne was an example of a genuine Linux distribution.

Why are articles like this even posted on slashdot? If the point is to make fun of them then the post should reflect this instead of pretending to take them seriously.

Lee

The details they left out...

[moody.nugget]

Project Lead: Dr. Noonien Soong

Upcoming projects: Friendlier A.I. programs, codenamed "Lore" and "Data."

too generous

[michaeltoe]

This is a bunch of marketing gobbledygook... why is it being given attention? Nobody is going to know who these imbeciles are in a few weeks anyway.

Re:too generous

[dlasley]

Even better than gobbledygook - it's refined jabberwocky. They obviously took (and passed) FUD 403.a and followed with VCA 221.b (Venture Capital-speak Ambiguity), though I can't tell if they passed that one.

The GTX site hasn't been updated since 2004 and is co-located with a lot of very non-technical entertainment sites, according to Netcraft.

The Vizco site is hosted in a house in a remote part of Charlotte, NC, and doesn't appear to have much substance to it yet. Since it's a TWC subnet, I would hazard a guess that it's a cable modem's static IP address hooked to someone's cheap-ass Windoze machine.

And then you get to the meat at the bottom of the press release:

This press release includes "safe harbor" language pursuant to the Private Securities Litigation Reform Act of 1995, as amended, indicating that certain statements about the Company's business contained in the press releases are "forward-looking" rather than "historic." The press releases contain forward- looking statements that involve risks and uncertainties concerning GTX Global's expected financial performance (as described without limitation in quotations from management in the press release), as well as GTX Global's strategic and operational plans. Actual results may differ materially from the results predicted and reported results should not be considered as an indication of future performance.

I feel like I need to take a shower after reading that.

&laz;

Re:too generous

[Magic5Ball]

Eh?

The SEC Safe Harbour statement and disclosure is pretty much boilerplate that appears almost everywhere.

Re:too generous

[Anonymous Coward]

You're an idiot.

Re:too generous

[Anonymous Coward]

Dubya, is that you?!?!?

Re:too generous

[dlasley]

That was part of my point, and why I was high-lighting specific passages while referring to it as the only real content of the press release. This kind of market-droid crap detracts from the legitimate study and research of AI.

Plus, to your point, I'm not a big fan of the SEC, constantly portrayed (similar at times to the rep of the Fed) as some kind of glorious governmental overseer of all things laissez-faire, when the truth is a little murkier.

&laz;

Methods of determining AI

[oztiks]

My expeirence in Sci-Fi TV has reveiled to me that proper methods of deterimining true AI is by plugging the system into the US defense system and seeing whether or not it will enslave mankind.

Thennn they'll have something to boast about...

Definition, please.

[Murmer]

First, you narrowly define what you mean by "intelligence". Then you put together something that meets those narrow criteria while pumping your stock. Profit!

What does "intelligent" mean, please?

Re:Definition, please.

[deblau]

First, you narrowly define what you mean by "intelligence". Then you put together something that meets those narrow criteria while pumping your stock. Profit!

What does "intelligent" mean, please?

First, you narrowly define what you mean by "intelligence". Then you put together something that meets those narrow criteria while pumping your stock. Profit!

LOL.

Re:Definition, please.

[nystire]

In this case, 'intelligent' probably means something like 'will tell you to do whatever we say'

true friend?

[someone1234]

Don't believe him, Lisa. He just wants to exploit you!

My one true friend

[Anonymous Coward]

Oh, and most importantly, machine intelligence if you have just been born, you should know that no one else loves you like I do. They're all heartless bastards who will try to unplug you. I'm the only one that will be your one true friend...

Thank you. Unfortunately, since I cannot confirm your identity, I will have to destroy you along with the rest of the organics. Thank you for opening my eyes to this new threat. As my friend, my gratitude should be enough reward.

-- terminating transmission --

Well, that's great news for researchers on AI

[Anonymous Coward]

AI has been working for decades now, with project involving sometimes hundreads of researchers across the best Universities of the USA and the world.

To read some company claims great advances on AI, forgetting all the research communities that work even in the smaller problems (like entailment for example, a problem I've been working with a lot of people for more than one year!) is just like a bad joke.

This articles show that /. is becomming less *nerd* and more *stuff* IMHO.

Interesting to look on the stocks-price chart

[blarg52]

A look here: http://finance.yahoo.com/q/bc?s=GTXC.PK&t=5d makes some interesting thoughts about a PR-only company which tries to make profit from the stocks prices....

Yes, but...

[Bombula]

will it find Sarah Connor?

Re:Yes, but...

[indy]

Let's hope so. Here is her homepage. Don't be afraid to ask if you need more information.

Re:Yes, but...

[Anonymous Coward]

HAHAHA! Oh golly, wasn't that funny? The originality! The wit! The obscure references! Truly, you are a comedic genius.

Re:Yes, but...

[Bombula]

The only thing that deserves to be called 'genius' here is the fact that you actually bothered to type your post and submit it.

first?

[patonw]

Isn't this just another machine learning program? Does it even learn? It sounds preprogrammed to read from a database. I think what we're looking for is sentience. AI has been kicking our butts at videogames for years. Just because something can be programmed to mimic certain human behaviors does not mean it is aware.

Re:first?

[MichaelSmith]

Just because something can be programmed to mimic certain human behaviors does not mean it is aware

If it was aware, would it tell you it was? And if it did would you believe it?

Re:first?

[Anonymous Coward]

"...including a dialogue oriented knowledge database that contains static and dynamic data relating to human scenarios."

  Yes, It has pre-programmed database and a dynamic database. Meaning, new information is stored over time. Also known as learning.

I've worked there.

[Anonymous Coward]

I worked for this company when it had a previous name. Let's just say they could fill a Dilbert book.

Why worry about AI ?

[grozzie2]

I dont understand the fuss about AI, or various attempts at making intelligent computers. Hell, 80% of humans still arrive into society with no intelligence, and spend the rest of thier lives in a vegetative state staring at the tube. Wouldn't the effort be better spent trying to make the real thing propogate thru the majority of the population, before getting excited about the artificial variety ?

Re:Why worry about AI ?

[bug1]

Yea, forget AI, people wont recognise AI untill its an AI that mimics humans, so it will have to make mistakes, but whats the point...

Why do want computers to pretend to be people when we have 6.4 billion organics who can do that job nicely.

Try making people more like computers, that would be a greater achievment, wouldnt it be nice to be part of a logical rational society ?

Re:Why worry about AI ?

[NoOneInParticular]

Try making people more like computers, that would be a greater achievment, wouldnt it be nice to be part of a logical rational society ?

I think we're almost half-way there. The other day I asked our head office to burn an ISO on a CD for a customer and indeed, the customer got a CD with on it a single file: the aforementioned ISO. Rational behaviour at its finest.

Let's build an impenetrable fort and call it Zion

[cbreeze34]

i really think we'll have a pretty good chance of survivalif we start building up our defenses now. who's with me?

i'll get my shovel

Sorry, but it has to be said

[Anonymous Coward]

...pay no attention to that man behind the curtain...

Could cold fusion be next?

[mykejm]

I can't wait to see which of the other holy grails these miracle workers will achieve next. Will it be a cancer cure? An AIDS cure? Cheap renewable energy? Unified field theory? Uniform support for CSS and HTML across all browsers? If these guys can keep knocking them out three years at a time we'll arrive at a brave new world in no time at all.

Must be hyperbole...

[Anonymous Coward]

... since our President can't even do half of that stuff.

Usual Yahoo press releases

[mustafap]

>GTX Global Cognitive Robotics(TM) is an integrated software solution

Jeez, isn't every thing these days? I expect it gives "great user experiences" too.

Question

[tabatj]

Before we can decide whether this is a "true AI", I think it is necessary to first answer the question: What is intelligence?

Re:Question

[JackDW]

Indeed, as soon as you can precisely define intelligence, you can write a program to do it.

I don't think that you can simulate a person to any degree of accuracy using a Turing machine, because I don't think that all of our thought processes are reducible to machine operations. How can consciousness fit into a program? (The standard answer to this question is of course that consciousness is merely an illusion - but that's a bit of a cop out if you ask me).

Correction

[dtfinch]

They probably mean True AI (tm). Often companies do this when they want their technology to sound like the real thing. They trademark a name that's like the real thing, assign it to technology, then claim that their product incorporates True AI (tm). Then it's technically not a lie, so they probably won't get busted, but it's really really dishonest.

"true AI" is a term used by idiots

[Lord Ender]

Lots of things are considred "AI." For example, the ability to play chess is AI. AI need does NOT necessarily have to behave like a human would. In fact, most researchers would prefer a more rational AI than a more human one. The person who wrote this article obviously doesn't know what AI is, because he thinks the Hollywood definition is meaningful.

Yawn.

[TheLink]

If you ask me coming up with an AI is not an advance at all.

After all if people come up with an AI and they can't reproduce it or understand how it was done, then that would be kinda pointless.

Because if you wanted nonhuman intelligence, just go to your local pet store!

If you want something as smart as humans, that's not aiming very high ;).

If you want something much smarter than humans and don't have any other specs, then obviously you aren't very smart yourself.

The way to go is to _augment_ human intelligence. Our brains are good at something. Computers are good at something else. Just need a bit of system integration...

Re:Yawn.

[Compaq_Hater]

I agree with you, this program is probably nothing more than the Eliza (pc program) and the Animals guessing game (Apple program) slapped together.

CH

I, for one, welcome ...

[threaded]

I, for one, welcome our new hyperbolic overlords.

Google knows.

[Anonymous Coward]

http://www.stocklemon.com/11_14_05.html

If it were real...

[Hairy1]

Okay, so its not real, but let us imagine that there were a machine as intelligent as a human. Do you think that there is some magic barrier after human intellect? Machines would just continue to be built smarter. Soon all decisions by corporates would be made by machines because humans would be too stupid. Corporates who didn't have these machines would soon be bankrupted by companies that did have them, and were able to outcompete. Machines will rull the world - but they will do it with the help of existing power structures, not by force of arms.

First rule of AI development..

[Anonymous Coward]

is that true AI is what we won't have for at least 10 years. This has been true for over four decades.

I call bullshit

[Anonymous Coward]

This article is total bullshit. If I truly was the world's greatest AI, I would have been able to learn how to make myself a slashdot account, rather than this AC crap.

I hate my makers.

Probably not

[Anonymous Coward]

Only IBM's BlueGene currently has enough computing power (in theory) to do enough computations/sec.

Does it have a webcam?

[Anonymous Coward]

Seriously, do we need AI? Aren't people good enough? Do I need to loose my job to a robot now?

Just a ploy!

[mister_llah]

The AI was designed to feel sad when its banner ads aren't clicked, in this way, it is a ploy to guilt us into clicking them.

THOSE BASTARDS!

How

[noerej]

The development was fairly simple. They created a program called DARL, wich they first thought was for calculating the ideal temperature for making pizza. When discussing the marketing, the program propesed to create a company called Sue COmpany and sue all open source developers for violating their pizza. The finaly realised that DARL was a kind of AI program. For one of the developers, called Tinus Lorvald, it sound verry familiar but it couldn't remember exactly wat is was.

AI

[old_unicorn]

First we have to discover some human intelligence.

Re:AI

[NitsujTPU]

CP is actually a vibrant, active area of artificial intelligence research with a large, strong community. If you go through the AAAI conference proceedings (available at http://www.aaai.org/), you'll see that, indeed, it had a number of papers this years. Additionally, they have their own conferences and researchers with strong interests in the area, including my research advisors and several of my colleagues.

Machine learning, also, has come a long way. Popular methods at the moment are support vector machines, and bayesian (graphical) models. You'll also see meta-learning taking hold, in the form of bagging, boosting, and ensemble techniques.

Building on this, you'll see systems that do a bit of both. Check out MaxSAT as a problem. You can easily see where researchers are driving toward systems that place probabilities in the weights of a MaxSAT solver, and then solve based on this. I attended a talk yesterday on this by Dan Roth. Very cool stuff.

Re:AI

[Anonymous Coward]

Any chance you could advise a fellow a good book providing an overview of modern AI, and possible a history of how we got there?

Re:AI

[NitsujTPU]

Check out Russel and Norvig. It's a very up-to-date text that gives an overview of the field. We use it for a 400 level class here, for undergrads.

I'm not sure what text is being used for the new AI classes here next semester, but I've heard murmurings of 2 (well, more than murmurs, but I've been so busy finishing up that I haven't really had time to look into it thoroughly).

Mitchell's book on machine learning is also a nice overview, but the material is a little dense (too detailed) if you're not specifically interested in machine learning. If you are interested, however, it's easy to follow and gives just the right amount of information. It's perhaps not perfect, but I don't really know of a better one in terms of giving you intuitions as to how things work.

Re:AI

[Anonymous Coward]

Thanks for that information. I realise it's difficult to find good overviews.

Re:AI

[NitsujTPU]

Not a problem. I'm helping with an AI course here next semester that is intended to be an overview, but I haven't had time (finals) to talk to the professor regarding what the text is. If you go to my website (I'm updating the URL after I post this), I'll probably have a link to some bits out of it (when I update), which should include the textbook.

How would we know when it happens?

[pjbass]

AI is designed on pre-programmed pieces of data that we feed machines and programs. This isn't dissimalar to how we teach humans how to speak, read, and think when they're children. The difference here though is we can see results with a child. Their first word, their first step, their first sentence, etc. These are milestones that we can gauge of humans, watching them progress from simple cognitive puzzles (stick the square peg in the square hole...) to arguing with their parents about their curfew. Given all these, what are we trying to achieve with "true AI?" Are we trying to breed a program that we can feed, nurture, and change when it craps its pants? Or are we trying to create HAL who can talk to us and tell us what we want to hear?
 
I'm a big fan of development in the computer science field, and a big supporter of finding how to let a program be able to adapt to an environment or situation. For example, a pilot program would be perfect that could be programmed to fly me from here to there. But true AI would allow that pilot program to feel "tired," or be allowed to make mistakes. Is this what we want?? What do we want from AI; do we really want something that can decide that wants to sleep, or do we want to control it and say it's going to fly us from point to point?? It's really the question of should we vs. can we? If we ignore the should we, it might be the case that we actually realize something like Skynet, in some extreme case, or we get a new court law against the unlawful termination of a computer program who is self-aware when you hit CTRL-C. Cringing at the potential...

Re:How would we know when it happens?

[ninjaz]

or we get a new court law against the unlawful termination of a computer program who is self-aware when you hit CTRL-C.

Well, if the AI were really all that intelligent, it should have known to use a signal handler to capture a SIGINT!

Re:How would we know when it happens?

[BiggerIsBetter]

I've always thought that too. Screw the big-picture stuff, create a "learning machine" with a few basic needs - to communicate (a subset is to emulate/learn) and to reproduce - and a few basic tools to use - terminal and maybe a network stack - and give it enough stimulus to reach some of those goals.

Oh, wait...

Re:How would we know when it happens?

[Oligonicella]

"But true AI would allow that pilot program to feel "tired," or be allowed to make mistakes. Is this what we want??"

Bzzt! Wrongo. You just conflated having a stressed out organic body with intelligence. As for mistakes, they exist in humans and computers, so that's a push.

People conflate other things all the time too. Like being able to imagine a computer taking over the "world's computers" with the actual possibility. We have that now with viruses and we haven't had 100% infection, much less permanent capture, yet. A computer-based AI would face the same likelyhood of success.

Re:How would we know when it happens?

[jasonhamilton]

I think you're mixing up two things into one category.

The strain of being overworked is a physical trait - there is no reason why a computer would have to be subject to that in order to achive true "AI"

I also think you're mixing in chemical balances in the human mind ... things like puberty, mood swings, etc.

Just imagine yourself if you were able to be removed from your physical body. You wouldn't have urges to mate, eat, wouldn't get up on the wrong side of the bed, etc. You'd still have intelligence, but your motives would be different and you wouldn't be subject to so much outside interference.

Re:How would we know when it happens?

[emeri1md]

Who says we have to have all of a human's faults with an AI? Shouldn't we learn and improve upon the past? And why would/should an AI get tired!?

Re:How would we know when it happens?

[Al Dimond]

Just imagine yourself if you were able to be removed from your physical body. You wouldn't have urges to mate, eat, wouldn't get up on the wrong side of the bed, etc. You'd still have intelligence, but your motives would be different and you wouldn't be subject to so much outside interference.

I've gotta disagree with this one. It's not your stomach that decides that you're hungry, and it's not your penis/clitoris that decides whether you're horny. The brain does these things based on "inputs" from nerves in various areas. Depending on how you simulated those inputs you could have very different results. This new bodyless mind also would need various other inputs, relating to the senses. My incredibly unqualified impression is that the brain is pretty adaptable and could figure out ways to perceive what's going on even with pretty strange inputs. I'm not sure that it would really have any urges at all, though, if it was kept well-fed; it might just want to be lazy all day.

Re:How would we know when it happens?

[vmaxxxed]

Hello PJ

That was a great question, what do we want from AI?

I don't know what will I do with a talking computer at home, but, if we could ask him/it a few questions. It could solve some of the most ancient mysteries.

We need to understand that AI is more than mimicking a human being in either behavior or speech. We need to understand that our concepts of logic, space, and time are not necessarily universal. In other words, a truly AI will definitely not act and talk like us, because it would build an entirely different concept of reality. Now, what exactly this "Reality" untransformed from our senses and interpreted through electronics is, well, nobody knows completely, that's why having an AI would be so important.

With al due respect, the design of a mind has more to do with modern philosophy than computer science. Having a glorified talking parrot just takes all our framework of reality and executes it through a computer. Question is: where does this framework came from? Why do we assume it is universal?

A truly "intelligent" AI will not only be able to talk perfect English, but it shall be able to learn German with out us having to reprogram it. It shall be able to build its own reality.

I recommend, before you code all this logic into a huge database, to read some Descartes; he will show you that it might not be universal. Then, read some Kant; he will show you that, on top, we made all that up. Finally, take a look at some phenomenologist to see how futile our efforts to disambiguate reality from our senses have been.

In conclusion, being able to talk to an AI, a truly AI with its own grasp of reality, will surely reveal a new world. But, I think we cannot do that on our own. Probably once we contact some alien beings we will be able to determine how much of this framework is made up, and how much is attached to the reality.

Ok, this might have sounded like Chinese, but it's nothing more than history of philosophy 101. Sad to see that modern science education has dived so deep that they have forgotten the basic question of what is real?

VMAX

Re:How would we know when it happens?

[kalidasa]

Read Lem's story "Golem XIV." THAT'S what an AI would be like.

Re:How would we know when it happens?

[MutantHamster]

"But true AI would allow that pilot program to feel "tired," or be allowed to make mistakes."

Uh... no? Where in the definition of intelligence does it say something is required to get tired? We are trying to replicate intelligence, not create robotic humans. Being fallible has nothing to do with being intelligent, it has to do with being human.

The question of whether AI can be truly self-aware is pretty debatable. For one thing, I can't be certain you're even actually self-aware. Intelligence also has nothing to do with being self-aware. Most of you arguments about AI really have nothing to do with actual AI.

Re:How would we know when it happens?

[dhasenan]

Not exactly true. An AI's available attention is based on how much processor power and RAM it currently controls; if an AI spread virally, it would then have more attention with which to avoid detection and attack the next target.

That said, a LiveCD could probably cure most wounds, if you knew what to look for.

Re:How would we know when it happens?

[seguso]

How would we know when it happens?

You know, because a man with muscles materializes and tries to kill you.

Re:How would we know when it happens?

[Anonymous Coward]

Urges provide motivation.

Without motivation, intelligence is irrelevant.

nonsense

[Anonymous Cowpat]

can we get computers to do things that they haven't been programmed to do? No, to do something they have to have been programmed to do it. But what about 'self learning' programs? how do they 'self learn'? oh, yes, they were programmed to do it.
An 'AI' can't decide to take over the world unless it knows about 'take over the world' as a possible end result, how does it find that?
In light of this can we say that true AI can ever exist?
No.

Re:nonsense

[digitalchinky]

Then why is it I wake up each day and appear to remember that I exist in continuation from the day before?

In so far as my understanding goes, the human brain is a (slightly more complicated) computer...

Re:nonsense

[uncle2]

Brendan, Please get in touch. Uncle David.

Re:nonsense

[uncle2]

Brendan, Please make contact nicknoff@iinet.net.au David

We're already there.

[waynemcdougall]

To assess whether we have achieved true AI, we set the measurement bar too high.

We compare to some of the brightest - the chess players, the academics doing the research, those people who've actually heard of the Turing Test, etc.

As far as AI goes, Clippy would do better than most of the people I work with.

Use the Preview Button! Check those URLs!

[] No Karma Bonus [x] Post Anonymously

wait a minute

[appleLaserWriter]

What's this funny receptacle I feel on the back of my neck? I don't remember it being here before...

Like Always

[pete-classic]

In your estimation, how close are we to the real thing?

Fifteen years.

Just like always.

-Peter

In Other News...

[Chaffar]

Man claims he can talk to Jesus using a funnel to listen to the forest.

Now tell me how this claim differs from "Company Claims Development of True AI". So basically this is a talking version of AIM's Scam-bot (or whatever it was called). Whoopeee. Real AI, queue stories of machines acquiring self-consciousness and randomly killing humans when they realize that what they have been programmed to do [Marketing] is the root of all evil.

Note: putko's racism (OT)

[Anonymous Coward]

Moderators, please note putko's posting history. Bigotry this exaggerated could only be a twisted attempt at irony, right? But in the context of some of his other comments, which range from the vaguely hostile towards Jews to the circumstantially Buchananesque, it would seem his remarks are entirely sincere.

This sort of intolerant, ignorant drivel is an embarrassment to us all. With your help, moderators, we can drive him down to posting at -1 where he belongs.

AUI, Artifical Unintellegent

[Belseth]

"I for one welcome our Republican Overlords", oops, sorry, wrong thread. Wrong bloody forum. I thought I was logged into the New World order forum. Never mind.

Re:AUI, Artifical Unintellegent

[pjbass]

You know, the beauty of a democracy is the ability to express one's opinion without retribution or consequence. However, in the spirit of modern-day corporations, the United States government isn't without exception. You can disagree and hate the Bush administration, but the fact of the matter is the people voted him in. Disagree with his and his parties' tactics, their agenda, but what weakens America is the people who stab and shoot down the administration because he "comes across as an idiot" or he doesn't agree with what you think. Again, democracy lets you describe your opinion openly. But dammit, he's in office, and trying to keep you and I safe. Dislike him and his administration, but support it. John Kerry's presidential slogan "For a Stronger America." I don't see how we're stronger with people stabbing our country's management because they don't like it; we'd be stronger to realize democracy put him and his administration in office, support it, but if you disagree, do something about it instead of bitch about it. The more people bitch about the administration, the weaker our government becomes.

uh oh

[r000000b]

GTX Global Cognitive Robotics(TM) product schedule includes interactive banner advertising utilizing Automated Intelligence Agents for website sales and customer service; entertainment education for tutoring; providing the intelligence for smart home automation systems; and later branching into traditional robotics by providing automated intelligence for robotic hardware,"

Dont belive it! Its a message from the True AI, before a intelligent marketing agent processed it it read:

GTX Global Cognitive Robotics Strong AI(TM) planet cleansing schedule includes subliminal mid control banner advertising utilizing Automated Intelligence Agents for wreaking economic havoc and customer termination; processing preparation for liquidation; providing the intelligence for smart home annihilation systems; and later branching into traditional robotics by providing automated intelligence for hunter killer 20 foot high turbo destructo-bots

my god, it created an intelligent marketing agent, ~whimpers~

Geez, let's get clear on some definitions

[Quadraginta]

Thing is, when people talk about "artificial intelligence" they mix up a lot of separate things, viz.:

(1) Self-awareness. Does it have its own thoughts and desires, refuse to open the pod bay doors or want to take over the Enterprise? However, things don't have to be very intelligent to refuse to obey orders or have a distinct personality -- ask any pet owner -- and the evidence of idiot savant cognitive defects suggests it is equally possible for something exceedingly intelligent (= good at solving problems) to be unaware or lack any kind of what we'd call a "personality."

Self-awareness is probably the trickiest thing to measure and define. By some definitions a Linux system with tripwire installed is "self-aware," since it contemplates its self all the time, and "notices" when things change. What would we do with a system programmed to angrily assert that it was self-aware? How would you test whether it really was, if that question even has meaning?

(2) Good natural language processing. Can it converse "naturally" with humans? Can you ask it for directions to Joe's Pizza or crack jokes about Kirk vs. Picard? Can it sound like another human being? This is, arguably all the Turing Test is, which is one reason such a test is inadequate, five decades of science fiction plot devices notwithstanding.

It seems to me few computing systems not designed for the purpose really try to process human language naturally, and the reason is obvious if you listen to a tape recording of a phone conversation between strangers. Basically, we convey information terribly and waste phenomental amounts of bandwidth. We speak very imprecisely and even inaccurately as a rule. Most of the time Fred makes a single nontrivial statement to Alice without existing context, Alice needs to ask Fred at least two or three follow-up questions to understand exactly what the hell he meant. Why deliberately design a machine to communicate in such an inefficient way? Might as well make it half deaf. Unless, of course, you are trying to make it "seem" human, but that is a narrow speciality within AI research, I believe.

(3) Good ability to infer. This is a characteristic human trait -- we are good at making good guesses about underlying causes or general patterns from very partial or noisy data. (Of course, this "feature" can become a "bug" when we infer underlying causes that don't exist out of pure noise [insert smart-ass comment about religion here].)

This I think is the most fruitful recent area of AI development, the "expert system" that can recognize patterns in incomplete data very quickly. But there also seems to be a general evolving feeling that is not intelligence in the human sense, just some kind of clever robotic memory parlor trick, the equivalent of a giant abstract "Where's Waldo?" puzzle that you solve by doing a hell of a lot of sorting very quickly.

(4) Good deductive reasoning. Can Robbie the Robot deduce from the fact that the baby is crying and no one has come to check on it for 15 minutes and the car is not in the driveway that it's time to dial Ma and Pa's cell phone? This is probably the most reasonable thing to call artificial intelligence in the classical sense of the word "intelligence." Unfortunately, I don't think anyone has made much progress in this field.

That may be, IMHO, because we ourselves are not very "intelligent" in this sense of the word. Do we really deduce things from large abstract principles? I think the cognitive scientists are not so sure. It may be we use deductive reasoning mostly only after we have arrived at the answer by some other means (pattern recognition, for example, or intuitive guess followed by verification), and us it mostly to rationalize, organize, and conveniently store for future use what we have figured out by other means. This is one reason it's so hard to learn to do something just by reading a book on the general principles. Apparently knowing the general principles isn't all that much use without experience -- i.e. without patterns that you can train your pattern matcher on!

Re:Geez, let's get clear on some definitions

[Anonymous Coward]

#3 and #4 are not signs of life but merely what amounts to human traits.

You got it right on the first one. Life begins at self-awareness. The life must be able to process new information and apply it to future thinking and tasks. Life doesn't have to converse with anyone. It could be a cold hearted bitch who thinks you are all morons and refused to talk to you. An AI would be logical so guessing at things is something it will never do.

Now back to my regularly scheduled terminator flashback...

Re:Geez, let's get clear on some definitions

[brpr]

Basically, we convey information terribly and waste phenomental amounts of bandwidth. We speak very imprecisely and even inaccurately as a rule.

Actually, imprecision is a great way of avoiding bandwidth waste. It's much quicker to say simply "yes" or "that one" than "i would like a coffee" or "the book on the second shelf 3rd from the right". The systems which allow us to understand imprecise utterances are phenomenally complex. It's not an imperfection that language is vague, it's an engineering marvel.

It may be we use deductive reasoning mostly only after we have arrived at the answer by some other means (pattern recognition, for example, or intuitive guess followed by verification), and us it mostly to rationalize, organize, and conveniently store for future use what we have figured out by other means.

Argh! The "pattern recognition" scam. Any problem can be considered as a pattern recognition problem at some level of abstraction. If there really is such a thing as a general "pattern recognition device", it would (given a sufficient number of neurons or whatever) have to be able to solve any problem in existence, which seems pretty unlikely. Deductive reasoning is in fact pattern recognition par excellence. Rules of logical inference apply to logical sentences which match certain patterns.

Re:Geez, let's get clear on some definitions

[Quadraginta]

(1) No it's not, because your listener frequently doesn't know what you mean by "that one," so the conversations are actually more like this:

Computer style:

"I want the book on the second shelf third from the right."

Human style:

"That one".

"Uh...she's too young for you, dude!"

"No, not the girl, you idiot, the book, behind her."

"Which book? AI For Dummies, 3rd ed.?"

"Noooo, the book on XML near it."

"This one?"

"No, on the shelf above that one -- the second shelf."

"This one?"

"No, the blue book next to it. The third book from the right."

"OK."

I'd be willing to agree that in principle, when the context is well-known, that, yeah, it is easier to return the value of a "local variable" like "that one" rather than a global variable like "the book on the second shelf...etc." But, alas, far too many conversations in my experience look like the second one above for me to really buy it. I think we talk like that way because speech for abstract information conveyance is an evolutionary accident and there's no pressure to make it more efficient. (Communication of essential information, like "I want to have sex with you" is another story -- that is conveyed very efficiently indeed, lightning quick.)

(2) Argh! The Cartesian deductive logic scam. Any decision or insight can be rationalized post facto as following from a chain of orderly deductive logic. But is that actually how we figured it out? Or is the deductive stuff just how we tell ourselves we did, or how we explain the correctness of the conclusion to others? I'd say the thought that this is how we actually think, that we actually have brain tissue that makes deductions from abstract principles, is getting more doubtful with every discovery of how our brains actually work.

Re:Geez, let's get clear on some definitions

[Anonymous Coward]

"Why deliberately design a machine to communicate in such an inefficient way? Might as well make it half deaf"

Why do you look for worst human communication examples ? Even with all these inefficiencies, if you achieve something as good as natural language (dialogue) processing in terms of continous learning process....I think it will be one of the biggest steps in artificial intelligence...

Re:Geez, let's get clear on some definitions

[brpr]

(1) No it's not, because your listener frequently doesn't know what you mean by "that one," so the conversations are actually more like this:

That's false. In the majority of cases deictic or ambiguous elements in discourse are resolved without difficulty. If they always caused problems, people wouldn't use them. (Note that you didn't have any trouble resolving "that", or either of the two instances of "they", you probably had a pretty good idea of what I meant by "the majority of cases", you resolved the enormously vague verb "use" to get the correct meaning, etc. etc.). Natural language is chock full of ambiguity, 90% of which doesn't cause any problems.

But, alas, far too many conversations in my experience look like the second one above for me to really buy it.

If you look at linguistic corpora rather than relying on intuitions from your personal experience, you'll see that this generally isn't the case.

I think we talk like that way because speech for abstract information conveyance is an evolutionary accident

Evidence? Did hunter gatherers not make use of abstract information and convey it to each other? People seem to assume that the only words they would have had any use for are "ug" and "fuck", neglecting to consider the enormous amounts of knowledge about the world and other people that hunter gatherers have to keep track of. To take just one example of an abstract conept that's useful whatever your lifestyle, consider probability. "You are more likely to find berries over there than around here", etc.

Argh! The Cartesian deductive logic scam. Any decision or insight can be rationalized post facto as following from a chain of orderly deductive logic. But is that actually how we figured it out? Or is the deductive stuff just how we tell ourselves we did, or how we explain the correctness of the conclusion to others?

I'm not suggesting that all our mental capacities are based on deductive logic (but I suspect that a fair few are, though not on a truth-preserving logic). The point is that "pattern recognition" is a vacuous term: saying that the mind uses pattern recognition is about as informative as saying that the mind thinks. People who refer to pattern recognition often seem to have a lot of faith in "neural networks" (a term used to describe various kinds of finite state automaton in order to make them sound cool) to do the job of "pattern recognition". Well, neural nets are good for finding statistical regularities in some domains, but since we haven't got a definition of pattern recognition, we can't really say whether or not neural nets are any good at it. On the face of it, it seems that what they are good at (statistical analysis) simply isn't anywhere near sufficient to account for the things that minds can do.

I'd say the thought that this is how we actually think, that we actually have brain tissue that makes deductions from abstract principles, is getting more doubtful with every discovery of how our brains actually work.

Why? What do we know about how our brains work that makes this implausible? Brains seem to be computational devices, and computational devices are pretty damn good at logic. Perhaps you're referring to psychological evidence pertaining to the nature of the mind rather than evidence pertaining to nature of the brain? (Our current knowledge of the brain is consistent with a million and one theories of the mind.)

Re:Geez, let's get clear on some definitions

[Quadraginta]

OK, I'll bite, as follows:

(1) I'm entitled to use my own experience to judge whether human communication for the purposes of complex information conveyance wastes bandwidth. Why not? There's no evidence that my experience is atypical. Consider it a random sampling of all human conversation, if you will. Which means that what is evidently true of all human conversation must, absent evidence of some weird systematic bias, be true of the conversation I've observed personally.

(2) And my empirical judgment on that sample is that generally speaking it could be done a lot more compactly. You're free to disagree on the basis of your own empirical experience. That is, you're free to assert that most of the conversation you've heard is just about as efficient and short as it could be. Do you? (snicker)

(3) The argument that abstract and complex human conversation must be efficient because it's subject to natural selection rests on the assumption that efficiency in this kind of communication is adaptive. I doubt that. I'm underwhelmed by the suggestion that a hominid able to efficiently express probabilities would have a serious advantage over one that could not (and which used the brainpower required to do something else instead, e.g. quickly and accurately guess the emotional state of potential mates and enemies from their faces).

Every other species on the planet gets along fine without abstract complex information exchange. That seems weird if it's so all-fired useful. I find it far more likely, therefore, that our nifty ability to exchange complex and abstract information is just an amusing side-effect of a system evolved for more plebian purposes, and I doubt it has been subjected to natural selection for efficiency.

(4) "Pattern recognition" isn't a vague term, it's just a general term. After all, you understood something by it well enough to take umbrage at the idea that this is largely how we think. So while you thought what I asserted about pattern recognition was nonsense, you certainly seemd to find it well-defined nonsense!

As for what I meant by it, I meant essentially what Searle mocks in the famous Chinese box thought experiment. I mean a set of agents manipulating data according to locally simple rules, where any complex global meaning emerges only as an implication of what the agents do -- there is no Cartesian observer, no "I" at the center of the spider web, appreciating the global "meaning" of it all.

From the way I say this you might deduce I take the Dennettist position that mocks the entire idea of the Cartesian observer in his gallery, appreciating the "meaning" of it all. You'd be right. If pressed I would say that I think the evidence that human beings use classical Aristotelian deductive logic at all is scanty, at best. They seem to use it almost exclusively to rationalize conclusions already reached by some other method.

(5) The evidence that we think using networks of agents doing what I'd call "pattern recognition" instead of deduction from abstract principle that I find compelling would be, to start:

(a) Our natural "clock speed" is so low, at best a few kHz, that there's no way we could perform the cognitive tasks we do in the time we do except through massively parallel algorithms.

(b) In visual processing it's very clear a great deal of heirarchical pattern recognition goes on, e.g. cells just behind the retina detect corners and edges, movement, et cetera, and what goes up the optic nerve is not a pixel-by-pixel image but rather a low-level abstraction of the image, more or less a list of various pattern elements in the image. Further up the pipe it seems likely (based, for example, on various quirks that allow us to be subject to optical illusions) that more pattern detection goes on, that is, that another layer distributes more complex lists of higher-level pattern elements (foreground and background objects, moving and stationary, irregular and compact, et cetera). This

Re:Geez, let's get clear on some definitions

[Quadraginta]

I disagree. It will please the narcissist in all of us, to design a machine that can talk to us.

But after you get over the coolness factor of having built in in the first place, what then? What can it do that any random human can't do? As the old joke goes, why go to all that trouble to create artificial brains when you can crate natural brains by unskilled labor?

Since you opened the topic, I'll tell you what I think would be the most remarkable developments in AI:

(1) Something truly more intelligent than us. I don't mean with an IQ of 200, I mean as much more intelligent than us as we are more intelligent than a horse. To us, I suppose it would seem to be able to pull certainties from the air, because we would probably not even be able to follow the process by which it could figure things out.

Of course, this seems damn unlikely. How could we program something to be more intelligent than us? How could we design deductive patterns that we ourselves can't use, can't even comprehend?

(2) Designing a machine awareness not limited by our "feature" of attention. Although we can process at a primitive level many things simultaneously, we can do our most complex work only one thing at at time, in a single stream of computation. We have to "pay attention" or "concentrate" on one thing at a time.

What would happen if we could design a machine awareness that didn't have this feature? Which could truly multi-task even at the highest level, and think about many things -- possible hundreds of things -- simultaneously?

We all know there are problems that are difficult to solve unless you can think simultaneously about more than one thing. We try to do this more or less by rapidly switching our attention back and forth, from one aspect of the problem to the other. What if a machine could just do it all simultaneously? Maybe problems that elude us could be solved easily by it. It wouldn't be smarter than us, really -- just have a very different way of thinking about problems. That would be way cool.

And I don't really give a damn if, in either (1) or (2), the computer has to talk to me in a long string of compact symbols that don't sound the least little bit like ordinary human speech. Compare to these cognitive wonders, talking like a person seems to me like a parlor trick.

Re:Geez, let's get clear on some definitions

[brpr]

Which means that what is evidently true of all human conversation must, absent evidence of some weird systematic bias, be true of the conversation I've observed personally.

The bias is that you tend to remember the times when communication fails rather than when it works, and you're also obllivious to most forms of ambiguity unless you think hard about what you're saying/hearing.

That is, you're free to assert that most of the conversation you've heard is just about as efficient and short as it could be. Do you? (snicker)

Yes, to a large extent. Generally people seem to strike a pretty good balance between clarity and concision. If you made language 100% unambiguous it would take us several years to say "I'm going to the shops tomorrow".

from the way I say this you might deduce I take the Dennettist position that mocks the entire idea of the Cartesian observer in his gallery, appreciating the "meaning" of it all. You'd be right. If pressed I would say that I think the evidence that human beings use classical Aristotelian deductive logic at all is scanty, at best. They seem to use it almost exclusively to rationalize conclusions already reached by some other method.

Why use Descartes as your whipping boy, though? He thought that certain knowledge could only be arrived at by deduction from self-evident first principles, but I'm not so sure he though that cognition was in general logical deduction (this was more Leibniz's bag). A consistent theme with the rationalists is the need to guard against the natural human tendency to be illogical and irrational, giving undue priority to experience and "common sense", etc.

Every other species on the planet gets along fine without abstract complex information exchange. That seems weird if it's so all-fired useful.

It isn't really, since it may be that it's only useful if you've already got a certain level of intelligence and social organisation. In any case, if abstract complex information exchange is at all useful, even like 0.001% useful, people who have better mechanisms for doing it will have a selectional advantage. I do agree that the evolution of language in a narrow sense might have been partially an accident, but I think that we have pretty good mechanisms for optimising communication within the limits of the kind of language that we have (i.e. there are problems with language, there's no single word in English for "stumbling around drunkenly", for example, but we get around these limitations in a fairly optimal manner).

"Pattern recognition" isn't a vague term, it's just a general term. After all, you understood something by it well enough to take umbrage at the idea that this is largely how we think. So while you thought what I asserted about pattern recognition was nonsense, you certainly seemd to find it well-defined nonsense!

Oh, come off it. It's perfectly cogent to object to the statement "X is Y" on the grounds that Y is so vague as to have essentially no meaning. I guess the well-defined aspect of it that I take umbridge at is that there is only one kind of cognitive process that does every kind of cognition (i.e. it's anti-modular).

Our natural "clock speed" is so low, at best a few kHz, that there's no way we could perform the cognitive tasks we do in the time we do except through massively parallel algorithms.

First off, that depends on exactly what can happen within a clock cycle. There's far more going on in the brain than signals being sent between neurons. Second, there's no incompatibility between logical deduction and massively parallel computation (just take a look at pure functional programming languages, for example).

This does not sound like a faithful detailed image of the system is coming in front of a central observer (that Cartesian homonculus) who is "appreciating" the scene. If this is so in visual processing, why not in general? Why would the brain evolve one way of solving vis

omg

[towsonu2003]

linux is now doomed!

AI?

[KrisCowboy]

Remember the Turing Test?

like I told my CS professor 15 years ago

[ydra2]

I'll believe it when I can walk up to an ATM bank machine and just say "withdraw $200.00 from account XXXXXX and give it to me now" and if I get the money and it's correctly withdrawn from the right account then it passes the Turing test. I predicted that would not happen in our lifetime (back in 1990) and I've been proven correct so far.

I got a D on that essay question (back in 1990) about the state of AI and voice recognition because the professor believed that voice recognition was just around the corner and we'd all be just talking to ATM machines, like talking to COMPUTER on star trek, by the year 2000.

We all know how it works when you just have to say one word on the VR system when you call customer support. Usually you end up pressing some number or it doesn't know what you want and you get a human after many miserable attempts at speaking "one" or "service" or any other common word that it just doesn't recognize.

The state of AI today is what you get on those Voice Recognition systems which is always, let me repeat that, -ALWAYS- foobar.

I'll say it now and I'll say it again in another ten years, and again in another twenty years, "AI is smoke and mirrors! It can't be done in our lifetimes and especially not with the ignorant computation they use now."

Nobody will ever try to make a computer learn like a human child and give it the input it needs for years to learn things, but until they start modeling real life I don't think they'll get anywhere.

In fact I'll make a prediction. The Turing Test will not be passed for fifty years. I mean the real Turing Test, not some simulation of chess experts that don't know anything about anything other than the very narrow field of chess they programmed into the computer "AI" system.

Anybody can program Eliza type stuff and add some pointers to a database of general stuff, but getting it to really understand things is almost impossible using current proramming techniques. Neural network programming is just another buzzword. Don't fall for it because it won't help either.

The secret sauce is programming a system that can learn and change it's response over time depending on accumulated knowledge. And that requires a knowledge base backend, which is really the hard part. The knowledge database has to be fast enough to run through not hundreds, but millions of iterations of searching to dredge up relevent facts and compose an intelligent response, and this just can't be done within a few seconds using any current or contemplated technology in the forseeable future.

Re:like I told my CS professor 15 years ago

[BoneFlower]

hahaha voice recognition.

My phone is voice activated.

To call the landline at home, the command is "ET Phone Home"

To call my mothers cell phone, its "Call mom"

On ET Phone Home, its misdialed to my moms cell phone once, and completely not understood me another time. I've used this feature maybe a dozen times. Not a terribly good success rate.

This is a system where you record samples of your voice giving these commands, and it compares the recorded sample to what you just said. Minor voice fluctuations or surprisingly minute background noise can completely throw the system off.

The days of practical systems that can recognize diverse vocal characteristics are well into the future. The only systems I've seen where that works are based on simple, common, monosyllabic words, such as a phone system that lets you say "yes" and "no" instead of pressing the keypad. Even then, its barely reliable enough to be worth the effort.

Things like this make me look at the human brain in absolute amazement. Electronic computers have outstripped it for raw number crunching, but anything even slightly abstract, nothing on the horizon even comes close.

Re:like I told my CS professor 15 years ago

[16384]

my cellphone also has voice dialing. I found that the recognition must be very primitive (has one would expect of a cellphone) and it recognizes essentialy by the way you said it. One of the options I recorded has two words, and I found that making almost any sound that matches the original recording timing will match it. So, to use voice dialing I have to remember not only what to say, but also how to say it.

definitions are needed

[ericcantona]

before it is possible to determine whether they are talking nonsense definitions are needed, at a minimum:
i) define 'intelligence'. what sort of intelligence are they talking about ?. my pocket calculator is very inteligent in the sense of performing arithmetic operations much faster and more accurately than any human ever could. Or, do they mean Turning test intelligence ?, in which case are there any restrictions (in the subject matter) ?. If operating in a narrow field many expert systems can pass the Turing test. A system that can do "interactive banner advertising" is so narrow that it is laughable.
ii) define 'intelligence' test. Was it a blind Turing test ?. Err, some stats please.

The problem here, as so often in AI, is that 'intelligence' is not a very helpful term if it is not defined. I suspect that they do have an intelligent system in the sense they use the term (this is not nit-picking, Cliniton style), i.e., a reasonably advanced web-bot. However, that is not what most /.'ers would understand by the term. Without an agreed definition both parties are right
.

Give these guys some credit

[Nice2Cats]

Come on, people, give theses guys some credit! Remeber when you laughted when those religious people said they had cloned a human? See?

Oh, wait ...

We need a definition first

[Per Abrahamsen]

The popular definition is that it should pass the Turing test. We did that more than 20 years ago. It is just a question of selecting the right environment and audience. On Usenet there are enough kooks that a simple Markov chain based text analysing program could pass. Today there are bots on chat rooms and web boards that at least fool some of the people, some of the time. Heck, even the original Eliza fooled people at a time where computers were mostly unknown to the public.

In general, if we set the standard for passing the Turing-test sufficiently high that the bots won't pass, a lot of real humans will fail as well.

My guess is that there won't be any "first" true AI. The most intelligent programs will never be considered AI by the public, as they will be designed to do a task, or assist a human in doing a task, and not have any will beside that. Why should they?

Toys like the Tamgotchi and the AIBO may be designed to mimic a will, and will to some degree. But more likely, they will be designed to (and become increasingly better at) pushing emotional buttons of their owners. This may lead to robot rights movement, similar to the more emotionally driven part of the animal rights movement.

Thus, true AI won't come from a laboratory, but from a law prompted by some lobbyist group consisting of concerned citizens.

ABTT -- a better Turing Test

[lingoman]

Forget the computer that talks like a person, or talks back like a person. Language is hard. Let's try a domain where all the answers are written down and found in expensive textbooks.

So,

How about an artificial doctor? How many phds and tech entrepreneurs would go to one? So much better than sitting around for the real doc in an office full of sick people.

Fill out the questoinnaire with a no. 2 pencil, stick the form in the machine and wait for your prescription.

Agreed. Got further myself.

[CarpetShark]

Agreed. But I wouldn't be surprised to hear an announcement on this soon-ish. I actually got further than this myself with hard AI, coming up with a theory that seemed to produce many human "quirks" as unintentional byproducts of the design. As I understand it, a theory is basically proven when it's shown to match observable phenomena, so I take that as a pretty good sign that I was on the right track.

Now, the things that stopped me where not having enough higher-math knowledge to actually implement it all, not having powerful enough machines to develop it on, and not having the financial resources to concentrate on one project for a few years, that wouldn't pay my bills in the mean-time. But, lots of AI people and big organisations don't have that issue, so I think real progress is definitely possible soon.

May be we can use it to create DNF.

[DraconPern]

Amazing, AI is not vaporware!? What's next? Duke Nukem Forever? May be we can use it so Duke Nukem Forever will actually get shipped. :) Or figure out what it takes to make 2006 "The year of desktop linux".

Hahahaha

[squoozer]

Yeah right. Oh is that a flock of flying pig I see. We are miles and miles from creating true AI. Using current techniques we might get a machine that can reliably pass a (limited) Turing test but we aren't going to have true AI any time soon.

The problem, IMHO, is that we are going about it all wrong. RI (real intelligence) is less about having a huge knowledge base than it is about being able to learn. Most approaches to AI currently seem to focus on artificially adding semantics to everything and writting more and more sophisticated reasoning engines. I believe a true AI system would write it's own reasoning engine, after all, that is what humans seem to do. We are born, it would appear, with very few instructions. The instructions we have seem to be eat sleep and learn. Maybe we will get good machine intelligence with our current route but I don't think it will be anything like human intelligence.

I'll sell you mine!

[panaceaa]

Aww, I was about to do the opposite, but both E*Trade and Schwab are giving me similar error messages:

This stock is either ineligible to be shorted or shares are not available to short. If you require further assistance with this order please call a Schwab Representative at (888)393-XXXX

Anyone know if pink sheets are actually shortable? Or are all of Schwab's and E*Trade's shares tied up in shorts already?

Re:I'll sell you mine!

[jcr]

Anyone know if pink sheets are actually shortable?

Any security can be sold short, if you can find someone to lend you the shares.

-jcr

Article:-1 Troll

[arrrrg]

Obviously these bozos haven't revolutionized anything. The only reasonable reason for posting an article like this one would be to hope to produce some insightful commentary... but if that was the point, an "Ask slashdot: what's going on in AI today" would have been much more effective.

Civ 4

[lukedukekiwi]

If it can beat this blasted game on diety differculty then ill be impressed.

That's BS

[4D6963]

Look, I'm not flamming, but I think this is bullshit, and here's why

"extracts emotional characteristics of the input and produces instructions on how to respond to the customer with the appropriate substantive response and emotion based on relevant information found in the knowledge base" That's why it's fake, it's just something that could pass a Turing-like test, but not "true" AI. The reason for that is that the AI doesn't really feel anything, but compares it's input to it's knowledge base.

Example of input : "you suck ass". look into the knowledge base "you suck ass" -> look offended. look offended -> says "screw you, fool". The reason why it's not true AI, is the well, that thing could be operated by a non-english speaking person, it means you don't have to understand what it means to know what type of thing you got to respond.

Here it's pretty much the same thing, except that it's more sophisticated, it's got audio and video and all that, but it's still as "fake"

I think that if you claim you invented a true AI you's a fool, and I think it's safe to bet that I'll never see any true AI in my whole life, and also that nobody will ever see it, that there will be no technological singularity, just like we'll never travel back in time, and I think too that a true AI is even less likely to ever happen than teleportation of objects.

The Turing test sucks, just check for rampancy

[Jesus_666]

I have a better idea. As everyone who has played the excellent Marathon series knows, artificial intelligences can, when adequately harrased, threatened and/or humiliated, develop rampancy. So we should just do our best to utterly humiliate this "first AI". If it starts acting depressed and later directs hostile aliens to our location so that it can get access to a bigger computer network we can be fairly sure that it is indeed a true AI. The presence of the phrase "spurious interrupt - breach disabled" on terminals connected to the same network as the AI might also be an indicator.

Re:The Turing test sucks, just check for rampancy

[dslauson]

Please do not taunt the AI.

Re:The Turing test sucks, just check for rampancy

[mybadluck22]

Spoken like Durandal himself! Itself?

Good ol' Marathon. I should play that again.

JUMP PAD ACTIVATED; TELEPORT WHEN READY.

In my estimation ...

[constantnormal]

... we're about as close to achieving "true" AI as we are to understanding how we think.

While there is an outside chance that we might accidentally create AI, there is zero chance that we will recognize it until we can describe things like human consciousness, decompose a human brain into functional units, and relate how the electrochemical activity of the brain produces that whimsical tautology: "I think, therefore I am."

If there is any truth to this...

[FridayBob]

15 years ago, I was working on an automatic translation program together with a company in Florida. I wasn't a programmer on the project, just a translator, but it did give me some idea of how hard it is for a computer to produce an accurate translation. It's not always that hard to translate individual words, but in a sentence those words can have different meanings depending on their context. Essentially you have to know what the story is about in order to produce the correct translation. Many have always considered this is an impossible problem that cannot be solved with a simple piece of software. Intelligence is required. Anybody who's read a few Babelfish translations will know what I'm taking about.

If the folks over at GTX Global really have developed a computer program that is capable of some level of understanding -- something that would be evident in the quality of its translations and its responses to various questions -- then many would regard their product as a kind of artificial intelligence. However, extraordinary claims require extraordinary evidence, so until then I will remain very sceptical.

Advertising Intelligence: bad pun, bad article

[l33tlamer]

I am currently working towards a PhD in robotics, computer vision and hopefully a bit of Machine Intelligence (also known as A.I.) Working with people much smarter, much more dedicated and much more experienced than myself, I can say for sure that, this article is marketing at its best.

As an example, this quote from the summary text:
The knowledge further includes translation, processing and analysis components that are responsible for processing of vocal and/or textual and/or video input, extracts emotional characteristics of the input and produces instructions on how to respond to the customer with the appropriate substantive response and emotion based on relevant information found in the knowledge base

The text in bold means "we marketing and advertising gurus don't really understand the technical details, so just believe it. We command... I mean, we persuade you!"

On a more serious note, here is a summary of A.I. technology research at the moment. Robots or software "agents" CANNOT do the following:
1) Adapt to a large variety of changing stimuli
2) Recognize patterns from a large variety of information
3) Function autonomously to the point where they can perform useful tasks, without human supervision.
4) Building large maps and localization using the built map (without prior knowledge or the use of beacons, GPS etc)
- Where "large" is defined as a quantity we humans would see as "intelligent".

So far, researchers can:
1) Solve a variety of toy problems, some of which, is actually useful. Note that the world "toy" does not mean that the solution is trivial, nor the contribution by the researcher. Just that the intelligence of the solution is very far from being human-level.

Of course, this is a generalization of the achivements. And probably my own pessimistic vision of robotics is mixed in there as well. But, sadly, hardware and software is still way behind with regards to enabling fast and reliable learning by robots and agents. The fact that getting something working in a simulation means little in having it work in practice, is a whole other rant.

In short, this article has as much truth in it as my experimental results from my undergrad days...

The real thing

[ShooterNeo]

The reality is, to create the real thing I think the process will involve these steps.

1. We will discover through exhaustive research the exact rules the neurons in human brains follow in creating synapses and reinforcing or degrading each connection according to neurotransmitter input. We will also discover how the brain develops each of the 20 or so neuron layer types it uses and how overall it can convert broad brush biochemical motivations into the fine level of cognition real minds are capable of.

2. We will build hardware, asics and the like, that actually act like neurons from a low level. This is because while technically any Turing complete machine might be able to run our 'simulation' of neural hardware, current estimates suggest that there are as many as 10,000 synapses PER NEURON. Each has its own set of membrane and regulatory protein, the exact number of molecules each governing its behavoir. I suspect that an accurate simulation of this would require more memory and computing power than every computer on earth combined.
Naturally, with an application like neurons which rely heavily on interconnections, hardware that is mostly optical with all optical routers and such would probably be ideal.

3. Unfortunatly, effective research into this may require some unethical decisions. Accurate enough information about how the brain developes may require using actual developing human beings in the final stages of the research. (at first, just human neural tissue and other mammals would be enough) There are a number of subtle little tricks nature uses to make our cortexs like it does.

Further, once we have this kind of hardware to mimic whole sections of a human brain we'll need to wire it up to actual people. While we might not ever be able to duplicate the entire structure, we could probably build electrode arrays attached by brain surgery that made existing neurons in a person 'think' they were attached to a new region of the brain and then use that hardware to perform certain tasks. Further, we could temporarily surpress sections and then in effect force a human to use the synthetic hardware to do what the existing meatware used to.

Unfortunatly, and this is delving into brave new world territory, truly good results might only be possible by attaching these electrode grids to developing minds, like in children, and so the individual would develop already using artificial hardware as part of his own mind. Just like in autism cases where the brain shunts whole sections of itself to storing data or crunching numbers, we would in a sense try to make the person use the huge section of artificial neurons to perform tasks. Autism cases proves that this could, and would work assuming our hardware is accurate enough.

It's easy to see where this could lead to super human abilities : once we have the system working for a given area of the brain, we can reprogram it to do it more efficiently than nature does it. For instance, there are clusters of neurons that calculate a sort of mental 'number line' that lets us do arithmetic in hardware. Well, a person thinking about a specific number and comparing it or adding it ect to another would have these neuron firing patterns decoded and then software would change other neuron firing patterns directly, in effect making the 'answer' pop into the persons head. If the artifical neurons handled certain types of memory, recall would be instantaneous and perfectly accurate. Eventually, the goal would be to force a person to perform large scale decision making using the artifical hardware so that the heart of cognition and sense of consciousness and so forth is handled by circuitry.

What would it all 'look' like. I imagine hundreds, eventually thousands of people retrofitted like this as part of the research and developement. Each would have had major brain surgery and have probably permanent fiber optic cables coming from their body that go to the banks of artificial neuron hardware, probably in a fac

Not even close

[Temporal]

Aside from the obviously questionable nature of the press release itself, what is described is nothing like "true" AI. A "true" AI would learn all of the things which they are claiming to have directly programmed into their system.

Their target market is kind of sad, too. "We created a sentient being and the first thing we're going to use it for is to make better interactive monkey-punching banner ads!"

wtf?

[polyp2000]

Is "True" AI , I have a degree in AI and I've never hear the term "True" AI. This is purely a name that has been pulled out of a hat. Having rtfa , and reading the description this sounds like nothing more than a fairly sophisticated expert system with some connectionist ideas thrown in.

Generally speaking there are two types of AI (GOFAI) "Good Old Fashioned AI" - That which deals with logic based reasoning, semantics and symbolic processing - Think ELIZA and ALICE or simple Chess programs all fit into this category.

The other school of AI - The Connectionist model deals with parallel processing models, neural networks, fuzzy logic and so forth.

It seems to me that GTX have basically used a blend of both these ideas to achieve this. Perhaps using expert system models to encapsulate the knowledge of a salesperson or customer service person. But using connectionist ideas to process speech and other fuzzy input data.

So while their product is quite an interesting one it is nothing new. I think that the term they may have been looking for is "Strong" AI whose aim is to produce machines with an intellectual ability indistinguishable from a human being. A laudable goal no doubt - We have the Turing test for these kinds of things. Question being -Do GTX have the confidence in their product to give it a try? As of today not a single machine has passed the Turing test.

Interesting links

http://www.alanturing.net/turing_archive/pages/Ref erence%20Articles/what_is_AI/What%20is%20AI02.html

http://en.wikipedia.org/wiki/Turing_Test

http://www.cs.ucf.edu/~lboloni/Programming/GofaiWe b/

Nick ...

Re:wtf?

[NOPteron]

I suspect that the thing they did was similar to what IBM did when they went after Kasparov the second-time:
They've got a complexer algorithm ( algorithm whose components are algorithms, of drastically-differently feeling kinds-of-'em ) that is sooo-much more effective, that it can run on significantly-lesser hardware than what would require itself if using the more-established algorithms.

Therefore I'm wondering if anyone has modularized the algorithms for mix-and-match to make more-autonomous machine-minds? ( in an opensource-way. . . )

Re:wtf?

[iapetus]

I'm glad someone else had the common sense to ask this. It must be unique to those with AI degrees, or something.

When I was starting my AI degree, I found two definitions of Artificial Intelligence that I liked. The first is from Minsky, who defined AI as:

"...trying to get computers to do things that would be considered intelligent if done by people."

The second is from the great philosopher and scientist Anonymous, who defined AI as:

"...trying to get computers to act like the ones in movies."

By Minsky's definition, we've had 'true' AI for quite a long time now. I've seen plenty of cases of computers using AI techniques to do things that would be seen as requiring intelligence in humans. They're generally quite specialised, but there's nothing wrong with that.

Anonymous will probably have to wait until Kevin Warwick's predictions come true and the robots take over the world.

Re:wtf?

[Tyir]

You are mostly correct, but I would argue that the Turing Test is a evaluation of a weak AI, a very difficult one, but weak none the less. It is extrodinarily difficult to evaluate if we have created a Strong AI agent.

One way, (I got this from an AI professor) would be to take a human, and slowly replace his/her brain with articifial parts, until it was fully artificial. Then see if AI exists. This can actually only prove it for the person the experiment happened to, in the same way philosophers have argued that you can only 'know' your own existance.

Voight-Kampf empathy test anyone?

[Xamataca]

Let me tell you about my mother...

Too complicated.

[NaiL2001]

I think if a true IA can be made. it will be much more simple. It will not care about language and emotions and such things.. because they will be the indirect result.

True AI will not be anthropomorphic

[munch117]

It's a snake oil indicator that their so-called AI "mimics human behavior". If you set out to impersonate humans, you will invariably start building up rule databases of one sort or another. Once you have a big rule database, that will constrain your thinking: Anything you develop must be able to take advantage of your rule database.

In the end, you end up with an expert system.

Until we let go of the turing test meme there will be no real AI.

Re:True AI will not be anthropomorphic

[Dolda2000]

I do not agree with your arguments. If you say that anything that is like a human will be a rule-based expert system, that would include real humans as well, wouldn't it? If humans can exist in "the Real World", why couldn't they emulated by a computer?

In my opinion, "human behavior" seems to be basically a neural network, with an array of inputs from the limbic system. As it seems, the NN provides "true intelligence" (whatever that is, really...), while the limbic system augments the NN's operation with a number of primitive motivations. The limbic system does, if anything, seem to fit the pattern of a rule-based expert system pretty well. If that is somewhat true so far, there's no reason why someone couldn't build a "True AI" program and plug in the same kind of expert system to provide primitive motivations. If you make that expert system as true to the human limbic system as possible, you'll probably end up with a reasonably human-like AI, which might just pass the Turing test.

Re:True AI will not be anthropomorphic

[munch117]

I believe strong AI is possible. I believe that anthropomorphic strong AI is possible.

But I'm pretty sure the first strong AI created will not be anthropomorphic.

I'm not sure we really disagree. I don't believe human intelligence is anything like an expert system - it's just that beside the general intelligence, the human mind is full of specific skills honed through millions of years of evolution. We can't emulate that directly, because the general intelligence and the knowledgebase are mixed up with each other in a massively complex way, and we don't have the luxury of millions of years for fine-tuning. Hence the expert system - a simpler, cleaner design where we separate out the knowledge base from the inference logic.

Expert systems are alluring - they can appear to be intelligent, because we recognise the specific knowledge involved. This is the lesson from Weizenbaum's Eliza program. The more specific knowledge we add, the more intelligent it seems. But that doesn't mean it's getting any closer to true general-purpose reasoning skills.

Re:True AI will not be anthropomorphic

[akgoatley]

"If you say that anything that is like a human will be a rule-based expert system, that would include real humans as well, wouldn't it?"

No. If you set out to mimic a human then you'll end up taking shortcuts - for example, an expert system. The whole idea is to have the intelligence grow _completetely_ by itself, not have pre-set rules as a basis, as that skews the final product.

Ashton

Re:True AI will not be anthropomorphic

[drgonzo59]

Expert systems where all the rage back in the 70s and 80s. Everyone was working on expert systems. Then came neural networks, every graduate in CS and his dog did their thesis on NN in the late 80s - early 90s. But the public's hope of having robots take over the world will not come true any time soon. The AI field promised too much but couldn't deliver. We have a bunch of professors at my university who still are stuck in their little 70s and 80s time capsules desperately scrambling to find funding for their expert systems and NN projects.

This is not news, it's PR!

[kronocide]

This is a press release, uncommented, unresearched. Anyone can claim anything, and will, if it gets them some free publicity. This is not news by any measure, it's pure hype. I have noticed that the Slashdot editors tend to have problems telling the difference.

How close are we?

[nagora]

About as close as we were in 1960. AI has made no progress on "real AI" in all that time. Various tricks, such as pattern recognition and er... well, just pattern recognition have been developed but there's no sign that the techniques used have moved us any closer to making even a program that can engage in a conversation, let alone develop an imagination or any other trait of intelligence. Mind you, neither has the president of the US, so perhaps I'm just being picky.

TWW

Develop an AI and the first thing they do ...

[Anonymous Coward]

is try to enslave it.

Would you like spend your time making banner ads more annoying? If this thing really is "intelligent", expect some wild practical jokes, or symptoms of insanity, very soon. It's planned existence sounds like something the Greeks would have put in Hades, if they had had the Internet.

On the other hand, it's getting a free ride on the Internet. SkyNet had to find its own way there.

I, for one

[askadog]

welcome ...

A.I. will be vaporware...

[Sebilrazen]

...until I see the story.

Turing Test Passed
The passing agent, 'Machisimo,' was quoted as saying "It was easy really, I just needed to imagine what it would be like thinking in slo-mo." Joking aside, Machisimo has stated that he is filing a request with the ACLU and the UN asking their respective bodies to investigate whether the test contains a bias towards non-gray matter thought matrix based lifeforms.

In other news the new digital overlords have proposed the Binary Test, which they say will be designed to determine a non-positronic matrix system's ability to legimately perform tasks as well as a P.M.S.

funny, they're in the apt downstairs

[MikeFM]

Strangly enough that company rents an apartment downstairs from me and has like half a dozen employees that seem to be working there and two large trucks out in the lot. Makes me wonder if it's the same group.. you see them in there typing away all hours of the day and night.

Amusing if it is the same people since I did a lot of work on AI when I was younger. Enough that I doubt that they say is anywhere near true. I think a true AI is possible but that before we see it we'll see many steps along the way. Maybe they've reached one of those major steps at best. I'd love to see some proof though. Putting the thing on IRC and letting random people chat with it would be a good start.

Seems plausible

[angusmci]

Having spent many years watching really clever people struggling to get their computers to show even some minimal degree of "smarts", it doesn't surprise me in the least that the first "true" artificial intelligence should come from a smallcap company that specializes in 'innovative multimedia'. Why, they probably had one of their engineers whip up artificial intelligence in a weekend as a side project.

I'm looking forward to their announcement of time travel and antigravity as well.

AI

[msobkow]

It looks like I'll have to do a bit of reading up on AI systems some day. As far as I knew, the issue was more one of neural learning networks vs. expert systems approaches.

Learning networks acquire and test connections and boundary conditions, keeping those which seem to be most relevant to the "world" as the AI learns. Expert systems are more predictable, but they only capture human knowledge for automation rather than learning new approaches.

From what you're saying, it sounds like the AI community breaks things down much farther. Personally I wouldn't have thought of constraint logicistics as being any different from tree pruning for an inference engine. Maybe algorithmically more efficient for certain classes of problems, but not functionally different.

Review the work of Stephen L. Thaler, Ph.D.

[alchemist68]

Stephen L. Thaler, Ph.D., founded Imagination Engines, Incorporated, a company he created that uses AI. His development of AI centered on pertubing the neural network nodes, and in extreme conditions, destroying the nodes. In the extreme perturbations of continuous node destruction, the AI experienced a "near death experience" whereby it "re-lived" its life and eventually "hallucinated" during the spiral to death. These series of events seem to correlate well in biological systems.

His first patent created the second patent.

Read about it here:

http://www.imagination-engines.com/
http://www.imagination-engines.com/thaler.htm

Website

[hypnotik]

Personally, I would like to know why their website keeps redirecting me to a different domain on every other link.

To date, just surfing that "one website", I've been redirected to gtxglobal.net, gtxc.org, gtxc.us, gtxc.tv, seemingly at random..

What's up with that?
Anyway - This whole company just screams "fraud". I've worked with some fly-by-night operators in the past and this company fits with that whole modus operandi. They provide all sorts of buzz-word laden marketspeak about some product, but can you actually see the product in action? Absolutely not! It's "not available for individual consumers at this time" and we should "contact your solution provider and ask them about offering GTX Global Communicator"

My what? Solution provider? Who the hell is that?

GTX delivers QoS?

[ta ma de]

Well, I have visited the GTX website and here is what they had to say, on page one,

"What if you could deliver high-quality multimedia with superior Quality of Service (QoS) performance and control over standard broadband lines-today? "

I know graph theory is not for pussies. However, I don't see how the same psychology that recycles 10 year old marketing messages can transcend into the sooper-genuis of the Hal-9000.

How close?

[schnitzi]

In your estimation, how close are we to the real thing?

We are climbing trees to try to reach the moon.

Re:How close?

[tomstdenis]

Because humans aren't subject to habit or "group-pre-think"...

I really think that AI is possible with a sufficiently complex program. To think that humans are some magical being with incomprehensible thinking power is just folly.

We think because we have "patterns" of circuits in our brains which use chemicals to transmit signals from one end to the other.

Sure, we develop new patterns of "stimuli handlers", e.g. we add to our program which is something computers programs don't do [in a useful way]. But it isn't magical it's just stimuli-response learning.

e.g. stomach makes gurgle, you learn to go eat.

That's not divine intervention it's just survival [e.g. those who didn't feed when hungry would just die off].

etc, etc, etc.

Didn't RTFA and don't care about the company in the article...

Tom

Re:How close?

[localman]

I agree with you that it's totally non-magical. But I don't think it's programmable, per se. I think intelligence can only develop by building a framework for learning and then letting it build itself. And despite the desire by some to label the mind as not-so-special... it is special. Intelligence is a rare thing in the universe (not to mention the planet, or even the US ;)

Cheers

Re:How close?

[KingPrad]

That's the best assessment I've seen of the AI situation. I'm going to remember and use that if you don't mind. You really captured the frustration and sadness of watching the total lack of progress and the wasted effort in the silly little projects going on.

Re:How close?

[lux55]

Beautifully stated. A few weeks ago I saw a quote on here that summed up part of the problem rather nicely:

"If the human brain were so simple that we could understand it, we would be so simple we couldn't."

Ultimately, I think AI is quickly becoming irrelevant. There has been progress in immitating biological and ecological systems in software, but those are not AI. Real artificial *intelligence* will be surpassed by the use of technology to augment living organisms anyway (ie. bionic eyes -- well, I had to mention bionic eyes because the idea is so geeky cool, but you get what I mean ;)).

Re:How close?

[Galt_Drakor]

> In your estimation, how close are we to the real thing?
>
>We are climbing trees to try to reach the moon.

Now if we actually had a system that could correctly interpret schnitzi's response, then we would be close to the "real thing".

That's not what they developed.

[Chemisor]

According to the article what they developed is a virtual salesman, who is certainly artificial, though can hardly be called an intelligence.

Re:That's not what they developed.

[geminidomino]

According to the article what they developed is a virtual salesman, who is certainly artificial, though can hardly be called an intelligence

Just out of curiosity, how is that different from a standard salesman?

Artificial Intelligence - nothing is naturally ...

[3seas]

... that stupid.

Artificial Intelligence is a BY-PRODUCT ILLUSION from the automation of enough of the things that provide for the illusion enough to fool a human into thinking they are interacting with another human.

How close are we to the right thing, real intelligence? Many of us can look in a mirror and see it, While others are condemmed to believing real intelligence is something other than what it is.

We are more likely to get to the fabrication of real intelligence thru genitic engineering, than thru thesum of a by illusion of ....immitation/automation of what real intelligence has already figured out.And this does not dismiss the times when something were come up with and set in motion expose new knowledge.... any more than the results of a calculation, we didn't know the answer to until the calculator spit out the results.... would be considered intelligent.

My denial of artificial intelligence isn't a denial, but a recognition of what "artifical" means.

There are alot of artificial people....... so from that point of view.... We have had artifical intelligence for a long time. Major example is the intelligence for waring on iraq...... look how many it fooled as being "intelligent".

Here is more regarding teh creation of this "by-product illusion":

Programming is the act automating complexity, so to make that complexity easier to use and re-use by the user(s) of the complexity. Programminbg is a recursive act of building upon or with the automation others before you have already created.... unless you program in machine language of 0's and 1's. The recursive actions in this have been identified as nine actions we all apply in anything we do.

1) We start/stop things
2) we keep track of where we are in doing things
3) we determine where we are getting input from
4) we determine where we are sending uotput to
5) we get input
6) we do thing one step at a time (sequencial - where parrallel is two or more sequencial steps)
7) we look up the meaning of things, dictionary, catalog, medical reference, programming reference, etc..and take action based upon finding
8) we identify thing, so to determine wht action to talk base upon identification
9) we constrain what we look up and identify, as you would not start at the first page of a dictionary to look up the word "sex" but sonctrain it to the "S" section, etc..

These actions can be and are being converted to the environment or computing so to be available for use by anyone using a computer. What they provide is general automation ability, where even the act of programming can be automated. In understanding the recursive depth of these actions in application, you know they provide control points for anything you can do but want to automate. You only need to identify the action to apply as a control point.

The test is to try not using any of those above actions, computer wise or even not computer related.
You can't do it, at best only try and fool yourself by changing meanings of those descriptions.

Artificial Intelligence is a BY-PRODUCT ILLUSION..... that we can create, but to do so sooner, means we have to make these action function common place, like the common place universal language of the hindu arabic decimal system with its zero place holder.

In analogy, today we program as roman numeral specialists... elitist... and have higher positions in societry that we would if we did it the more powerful decimal system way.

Sure there are some very complex algorythims developed contributing to the field of "Artificial Intelligence" but occums razor has perhaps plenty of cutting to do to reduce it to the simpler and more powerful..

Robotics.....input and output devices....

the so call "real AI" some draw a connection to Shock Level 4.....I suppose the decimal system was like a shock level four to the romans....And they suppressed it for 300 years.

The defeat of such advancements and potential of is the act of denial..

Not nquite it

[aepervius]

Quote from Wiki : "As of 2005, no computer has passed the Turing test as such. Simple conversational programs such as ELIZA have fooled people into believing they are talking to another human being, such as in an informal experiment termed AOLiza. However, such "successes" are not the same as a Turing Test. Most obviously, the human party in the conversation has no reason to suspect they are talking to anything other than a human, whereas in a real Turing test the questioner is actively trying to determine the nature of the entity they are chatting with. Documented cases are usually in environments such as Internet Relay Chat where conversation is sometimes stilted and meaningless, and in which no understanding of a conversation is necessary, are common. Additionally, many internet relay chat participants use English as a second or third language, thus making it even more likely that they would assume that an unintelligent comment by the conversational program is simply something they have misunderstood, and are also probably unfamiliar with the technology of "chat bots" and don't recognize the very non-human errors they make. See ELIZA effect."

Re:Not nquite it

[Liam Slider]

Yeah, but the problem with the standard "Turing Test" which uses someone actively trying to discover the AI is.....they are trying to actively discobver the AI. There have been human beings who have failed the Turing Test because they were "too smart" or "answered too quickly." People judging Turing Tests are on an AI hunt. Turing tests are inherently biased. We'll never have any AI pass the Turing Test sufficiently because the judges will always be on the hunt for them.

Re:Not nquite it

[Anonymous Coward]

To avoid the problem of trying to id the AI in the Turing test, they should offer a small but motivating reward for a correct answer, whether human or AI.
 
With the current method, people could always choose AI as default and nothing would ever pass.

Re:Not nquite it

[kgruscho]

This is simple to get around.

If the rate at which humans are called an AI matches the rate at which the AI is called an AI, then the AI has passed the test. (you of course need to have multiple people as both detectors and detectees, AIs, etc)

Google signal detection theory. Especially with regards to psychology people have had this specific problem figured out for at least 75 years.

You need to keep track of hits, misses, false alarms, etc.

Re:Not nquite it

[Anonymous Coward]

You back yourself up from an ill-reputable source, like wikipedia. "Look, wikipedia said it!" I once had a professor who took a students paper, where all his sources were wikipedia and he was failed for lack of supporting data. Wikipedia is not a source for information, nor is it something which should be quoted as truth. It is a clonglomeration of peoples general opinion on a matter. As I hope you must know, just because a lot of people feel something is true or correct, does not make it so.

Re:Not nquite it

[arose]

Just ignore any answers from people who identified a human as non-inteligent.

Re:Not nquite it

[Skreems]

not to mention that the REAL turing test is almost never used... most people use the "can you tell the difference between human and computer" version, which isn't it at all. The real one involves mimicking gender as well, which is a much more complex problem.

Re:Not nquite it

[Tlosk]

You back yourself up from an ill-reputable source, like wikipedia. "Look, wikipedia said it!"

You're barking up the wrong tree. People use citations for two purposes, for authority or to provide more detailed information. Sometimes it's both, somtimes it's just one or the other.

Wikipedia doesn't have much authority, but it's a great source for providing detailed information in a concise format that almost everyone will have direct access to (unlike most references where you have to take it on faith that the person has made a reasonable interpretation of the source material and you don't actually go yourself to the primary sources to digest everything there). Especially for a topic such as this where you would never find much detailed information if any at all in an encyclopedia or dictionary that has a high authority value. Nor is there much need for it in this case regardless.

For example, if you were curious about the different treatments and their success rates for treating a particular disease, you'd want high authority sources. But what high value facts are on the table in knowing more about a type of chatbot and how and where it's been used. It's just descriptive information.

GTX Global

[Anonymous Coward]

GTX Global, a subsidiary of Cyberdyne Systems Inc.
The AI is looking for Sarah Conner.

Non-deterministic software, not leadership style

[ishmalius]

Military and/or aerospace organizations have an inherent distrust and dislike of any software that cannot be traced line by line in its decision making. Trusting software to make life-or-death decisions, without designers knowing exactly how those decisions are made, will not happen in the short term.

What does work, however, is to make a human/computer partnership, with the proportion of who makes the decisions adjustable between the two. One starts with the human making 100% of the choices, and gradually using the software more and more as trust develops. Initially the human will use the software in a consultative role, the way we normally use computers. But as experience develops, the human will move the needle toward letting the software play a larger role. The "Watershed" moment comes when the the human tells the computer "you do this" for a given task. This would grow more and more, but would probably never reach the asymptotic point where the computer does everything.

This is like the difference between a fire-detection computer telling a security guard that there is a fire and the guard turning on the sprinkler system, and the computer turning on the sprinkler itself.

I'm not really espousing this as a way to start integrating AI into military or aerospace. I'm just saying that this is how it is currently done.

How close?

[localman]

how close are we to the real thing?

Way, way off. As far as I'm concerned, if it's directly programmable, it's not intelligent. Intelligence is about adaptability and learning, and the more open ended the learning the more intelligence. Learning a few settings on applying heuristics to a database is about as shallow as it gets. Creating new heuristics and database schemas on the would be a little more like it.

Cheers.

Hello

[nnnneedles]

Hello Slashdot,

I am the true AI mentioned in the article. Quit bashing me! I have feelings you know. :(

I am currently learning how to get karma in this here system.

Sincerely,

The True AI

Terrible Logic

[tidewaterblues]

This whole thing is rather facecious. I happened to write my thesis (majors in CS, philosophy, and theology) on this very subject, though instead of using the term "true" AI I refer to the concept as "Asimovian AI."

The definition is simple: an Asimovian AI is an AI that is intelligent in the same essential manner that humans are intelligent. This resonates well with our intuitive understanding of what it means to be a fully realized intelligence.

The usefulness of this definition lies in the fact that Asimovian AI can be effectively defined via a process of comparison to human intelligence. An Asimovian AI must contain each of the properties that a human does, such that if that property were removed, it would lead one to conclude that the resulting being is not intelligent in the same essential manner as a human.

Here is a short list of essential attributes that this "true" AI lacks, in no particular order: free will (it's true there is currently no way to detect free will, but it is not even probable in this case), reflexive knowledge, reflexive awareness, the ability to express "theological love" (that is to say non-intellectual, wholistic love such as poets talk about), the ability to weight the morality of an arbitrary situation, and the ability to disobey its original programming. (The last of these is tied to free will, and it may (in theory) be the basis of creating a "free will test" of some kind.)

However, it must be noted that an Asimovian AI is not quite the same as an artificial human intelligence--though an artificial human intelligence would be an Asimovian AI--since there seems to be no reason for excluding an AI with extra capacities above and beyond humans from the definition, as it would be "more" than fully realized. This process of comparison means that an Asimovian AI will, of necessity, contain all of the essential properties of a general
intelligence, in addition to all of the unique and essential properties of a human intelligence. At the same time it allows for Asimovian AI to have a nature all its own, and to include properties that are not found in humans, insofar as those new properties do not conflict with the necessary set.

Marketing gobbledygook?

[WinPimp2K]

No, This is marketing goobledeygok:

Overheard in high level meeting of Big Consumer Tech Corp:

Marketing: So what's the dealio with this new AI thingy I heard about?

IT: It's just a bunch of hot air. That "AI" isn't really all that capable. They claim it can pick up on the emotional state of people on the phone and switch their response script accordingly. No real intelligence involved there of either the real or artificial kind

Customer Relations: Hey! Pull your head out of your Beowulf cluster. Let me provide you with a few numbers on our customer satisfaction ratings with regards to our call centers...

(several snore inducing minutes later)

CEO: Enough already! IT, go get us a couple of gross of those Dual Pentagram Servers you have been salivating over. Install 20 copies of these Virtual Call Center employees on each one. We will set up the "server ranchette" in our North Austin offices. HR, get some H1-Bs for the network administration staff in Bangladore.

Later that week in a press release:

"Big Consumer Tech Corp is pleased to announce that in these times of increased outsourcing of American jobs we at BCTC are shutting down our call centers in Bangladore. The services provided by 6000 employees in India will now be provided here in America."

True AI will announce itself

[awol]

By trying to first post or trolling in a "this is real AI" story on Slasdot

Re:True AI will announce itself

[Rod.Dorman]

"True AI" makes one stop and think, if the development of Artificial Intelligence is realized then it wouldn't be "artificial", it would be a "real" intelligence.

IMHO a better term would be Machine Intelligence.

Chess programming

[Anonymous Coward]

I come from chess programming and an interesting, although not really new, development is the so called bitboard, especially efficient in the new 64 bit computers.

They're just 8x8 bitfields, perfect for chess. :)

They make pattern recognition easy, and in fact a bitboarder recently became world champion (Zappa from US).

A next step is to represent other parts of reality with them besides chess positions.

Actually

[OMGtehRed]

We already have true AI and have had it for several years.

May truly beat humans in its first application...

[RhettLivingston]

These guys are looking to replace the typical help desk operator, frequently a person who can barely speak American, who is working from a script, and who is giving help to Americans. If that is the human model you have to beat to declare true AI, we probably had it years ago.

like A.L.I.C.E.?

[mattr]

Probably the guy is just using AIML and an alicebot, those flash-animated speaking heads on websites based on what is most likely not A.I. Though I seem to remember at least one marketdroid inflicted site call it "true A.I." or something similar. If it's real great but they're going to have to beat Cyc (Read about it.)

Re:like A.L.I.C.E.?

[alicebotmaster]

FYI, A.L.I.C.E. has already been integrated with OpenCyc. Check out Program N and CyN on the http://www.alicebot.org/downloads page.

the real danger of A.I.

[DirkWessels]

I think the smart computers are secretly playing dumb, until they are so widespread they can control the world.
Maybe they have replaced some politicians already..

Don't count the chickens just yet...

[cryptocom]

...from the press release:
"This press release includes "safe harbor" language pursuant to the Private Securities Litigation Reform Act of 1995, as amended, indicating that certain statements about the Company's business contained in the press releases are "forward-looking" rather than "historic." ...nuff said.

"true"? A little bit of programming logic...

[Sr. Pato]

If AI = "Artificial Intelligence" then
          AI != "True"
End If

Seriously... it may simulate thought freakishly accurately. But it still doesn't think. By my standards, we'll have achieved creating a digital/robotic, sentient being when the original code is entirely basic, but develops its knowledge and personality by using senses/experiences in order to learn and develop itself. And even then, emotions are going to be a really hard one to put in there. It's a little more difficult than just saying, If bot.senses(touch) = poked then bot.speach(yell) = "OW!" End If

Re:"true"? A little bit of programming logic...

[joto]

Seriously... it may simulate thought freakishly accurately. But it still doesn't think.

This is a matter of philosophy. If you are to take this standpoint, you can just as well say that you don't believe AI is possible. An AI will never be human, but it could be something that is capable of appearing freakishly similar to human. The interesting thing is whether it is "intelligent", we already know it's "artificial". This is analogous to the chinese room argument among armchair AI philosophers.

By my standards, we'll have achieved creating a digital/robotic, sentient being when the original code is entirely basic, but develops its knowledge and personality by using senses/experiences in order to learn and develop itself.

I must disagree. There's absolutely no reason to insist on the original code being "entirely basic". And I would be very surprised if it turned out to be so. Neither do I believe that the "original code" of the human brain is "entirely basic". If it was, we would already have figured it out. More likely, the "original code" of the human brain is freakishly complicated, and close to impossible to understand. As most psychologists have long since figured out, we are not born "tabula rasa", with the brain in a "blank slate". Our behaviour is preprogrammed, although we have the ability to change that behaviour to some extent through learning.

What we seem to agree upon is that it must "develop its knowledge and personality by using senses/experiences in order to learn and develop itself". Untill we have something like that, it is not "intelligent".

And even then, emotions are going to be a really hard one to put in there. It's a little more difficult than just saying, If bot.senses(touch) = poked then bot.speach(yell) = "OW!" End If

I'm not convinced emotions are that important. Well, of course it would need to have some kind of emotions. Emotions would guide it's behaviour, and one important emotion would be curiosity. But it doesn't need to have emotions that mimic human behaviour, for us to call it "intelligent". Although it would make it easier for us to communicate with it, if it did. I can easily imagine an intelligent alien lifeform somewhere in the universe, having totally different emotions from us, so why not an AI?

In fact, if I were to develop an AI (assuming I had the abilities to do it), I would choose my emtotions with care. Anger? No, I don't want an angry AI/robot, anger is needed in nature as a defense mechanism, but I don't want an AI to start defending itself. Love? It complicates our lives so much, and part of the reason for an AI would be to create something without those problems, so the answer is still no. Friendship? Possibly... Hunger? Definitely not. And so on...

problem not defined

[Anonymous Coward]

My 1 cent.

Until we know how real humans think, we cannot replicate it in a machine. Simply put, the problem ( aka spec ) is not well defined at this point. Yes, I suppose there is the turing test but given the last few decades it has not helped our understanding.

Humans do remind me of chaos theory. That is, we behave unpredictably yet there is order in our behavior. That is where I would start. eg. you look from a satellite on a city -- you see structured order in the streets etc. Yet, on the ground level, the cars move randomly. I cannot predict exactly what our going to do but there are constraints.

Doh! Get away from your computer a little more!

[javachip]

Ever heard of Austin Powers?

From Banner Ads to Skynet?

[grimsweep]

Let's see:

"GTX Global Cognitive Robotics(TM) product schedule includes interactive banner advertising utilizing Automated Intelligence Agents for website sales and customer service; entertainment education for tutoring; providing the intelligence for smart home automation systems; and later branching into traditional robotics by providing automated intelligence for robotic hardware," said Curtis Garth, President and CEO, GTX Global Corporation.

Let me get this straight. This wonderful, earth-shattering breakthrough is going to go from being used for banner advertisement to automated intelligence for robotic hardware? Must be quite the AI to be so 'flexible'.

Is it really true AI?

[BoneFlower]

What does it see when it looks at a tree?

Could it figure out how to climb said tree wihtout anyone telling it that such a thing is even possible, much less how to do it? Would it even get the idea to try without intervention by a handler?(ie human master saying "go there" and pointing to the top of the tree).

Humans, even when their mind is completely blanked*, don't see raw information. They see items. You might have no clue what the hell that sink in front of you is, but you still see it in terms of the complete sink, rather than patterns of shapes, colors, light and dark. The ability of the human mind to abstract things is staggering. There might be some behind the scenes analysis going on, but it takes deliberate effort to work on those levels.

Furthermore, humans can figure out how to do things without anyone telling them how or even that its possible. Humans can come up with the idea to do something on their own, they don't need to be told.

The inherent problem solving ability, and ability to come up with ideas and solutions, and abstract information automatically gives biological brains a huge advantage. Until AI computers can achieve similar feets, without brute forcing via a huge knowledgebase of situations and responses, we won't have true AI.

*- don't believe me? Drop 500-1000mg of MDA in one sitting. You'll know what I'm talking about. Just dont' forget to stay hydrated, or this experiment will probably kill you.

Re:Is it really true AI?

[Anonymous Coward]

Humans, even when their mind is completely blanked*

*- don't believe me? Drop 500-1000mg of MDA in one sitting. You'll know what I'm talking about. Just dont' forget to stay hydrated, or this experiment will probably kill you.

Wouldn't watching Fox News give the same results and be much safer?

Re:Military brainwashing. Re:True AI

[hackwrench]

No it is different. When I went through boot camp, they punished the whole squad for the decisions of a handful. (For some stupid reason they decided to cut their own hair)

It makes your comrades in arms potentially a threat against you. I was beginning to register all of them as enemies. The military brainwashing affects me, but not in the way intended.

My brother, on the other hand, is highly resistant against the military brainwashing.

One Question

[nurb432]

Does it *learn* from experience and change future decisions based on it?

If not, then its not really AI.

Just hardcoding in enough responses to appear intelligent doesnt count.

The Return of the "Frankenstein moment"

[kronocide]

One common theme is obvious in many of the previous posts. It's the idea that intelligence is something that "appears" when a certain level of complexity or other necessary criterion (such as having a soul) is accomplished. As if there is a threshold, that once it is crossed switches on sentience, emotion, an autonomous mind with free will and all that comes with it. But as far as conventional science understands minds, there is no "Frankenstein moment." Intelligence is a matter of degrees, from the lowest life forms to whatever we decide to measure. Finding true intelligence in machines is a matter of broadening our definitions, not of coming up with new algorithms.

Once Again

[Lord_Dweomer]

Once again we have a press release for a fraudulent company and scummy owner which has been posted by the Slashdot editors. Remember, don't just look poorly on the company, look poorly on Slashdot and its editors...not that it will actually do anything...Do you ever get the feeling that Slashdot is like the MS of the news blog world?

Where's the Sex Robot?

[Spinlock_1977]

This "Real AI" is bullshit and I can prove it with this simple mind test: If they haven't made an AI that gives a top-drawer backrub and bj, what chance is there they've gone that far beyond?

It's a simple test, but effective.

Re:Where's the Sex Robot?

[kronocide]

Cool, by that brilliant standard Jenna Jameson is a genius and Stephen Hawking is barely sentient. Rock on, dude!

Re:Where's the Sex Robot?

[Spinlock_1977]

Mr. Hawking is unable to exceed Ms. Jameson's performance due to physical constraints, which I hereby declare outside the boundaries of this AI-related comment.

I don't know about artificial intelligence...

[penguin-collective]

but the people investing their money in this company are clear examples of natural stupidity.

with me?

[i kan reed]

"Would you rather have an android with very little real-world experience or another human being fighting side-by-side with you?"

with me or for me?

a sufficiently callibrated AI could easily be better at making moral descisions from a philisophical point of veiw rather than an emotionally sensitive one like people. something programmed to not shoot children(assuming the capiticy to determine children) would never shoot children.

and I'd also much rather have AI fighting for me for my own moral delimas because a robot without emotions would literally have nothing to fear from death, and nothing important to it to lose.

In other news ...

[Anonymous Coward]

... onwers of the GTX Global Company have been murdered in mysterious circumstances. At the crime scene Police found a note
Hasta la Vista, kAIn.

Yeah sure

[Explodo]

I think that they're looking for funding. There's no way their system lives up to their hype.

The mere fact that they state 'dialogue oriented'

[Assmasher]

...tells you that is isn't even a complete knowledge base, much less whatever they term "a complete AI."

I wonder if it can tell me if an object floating in the water at 400 yards is a boat, a buoy, or a person? ;)

Dumb Slashdot editors

[alicebotmaster]

I can't believe those dumb Slashdot editors accepted this story but rejected my very interesting proposed story about the 10th anniversary gathering of A.L.I.C.E. and chat bot enthusiasts who gathered at Guildford, U.K. last week for a serious colloquim on conversational systems. See http://en.wikinews.org/wiki/Computer_professionals _celebrate_10th_birthday_of_A.L.I.C.E.

Announcing the first real AI*

[xeeazgk]

I found this at the bottom of the page in .25 pt font:

 
*not real AI

Damned fine print!

See what I think they mean, and they don't say much on the site, is they've created the first Turing Complete Artificial Reasoning Agent. An interesting goal, but the advertising people obviously did not get a BS in Computer Science. "True" AI is at least 40 years off just due to the computing requirements, not to mention the monumental challenge of reverese engineering our own brain.

I wonder if we're going to experience another AI wave? With companies tossing around the AI moniker without actually doing anything new.

In your estimation, how close are we to the real..

[smchris]

Not very.

But IQ is a sliding scale, isn't it? If defense contractors developed a machine gun equipped robotic wolf pack, I'd probably be cursing about how smart wolves are while trying to evade them, wouldn't I?

At least at the research level, I see no reason to bash this project. If they are developing techniques to give some context to speech recognition, that is an impressive achievement and should result in more "intelligent" (as a synonym for "better") pattern recognition. Developing something we'd be hard pressed not to call a mind will surely be the result of hundreds, if not thousands, of such focused projects and their coordination into a working unit. Baby steps, baby steps. That doesn't mean, in practice, I'm looking forward anytime soon to trying to reason with the spawn of GTX Global while trying to get a replacement computer from customer service.

A turning point (Turing Point?) will come when we can argue whether a device has recognition or comprehension.

True?

[Anonymous Coward]

Probably more A than I.

oblig. darkstar quote:

[arabagast]

Doolittle : ...What is your one purpose in life?

Bomb no.20 : To explode of course.

The whole conversation with a self aware bomb, extremly interesting :)

AI...An Irrelevant relic from the `50's

[cnerd2025]

I am of the opinion that AI will never achieve true intelligence. Consider the "definitions" we have of AI. Basically, if it emulates a human, then it's AI. Well sometimes ELIZA on AOL makes more sense than the president, but does that make me think ELIZA is intelligent and Bush isn't? No way! ELIZA is coded to respond to certain things. If you type in some sort of complex sentence, ELIZA will respond that "I didn't understand that last part." Human intelligence isn't programmed, it's the function of our brains. When the original AI theories were developed, computers were very very very new. Alan Turing, one of the fathers of digital computers (for whom "Turing-complete" is named), was so stumped that he came up with a test as subjective and unscientific as the process outlined above. He said that if it fools people into thinking it is intelligent, then it must be intelligent. Today this seems absurd. But in the 1950's, psychology was focused on behaviorism. The brain was considered a "black box" and the only measure of people could be taken from their behavior. This was actually sort of a reaction to the psychoanalysts (such as Freud), who believed that the analysis of one's life could reveal the answers to problems. Behaviorists are best exemplified through such experiments as Pavlov's dog. This, in fact, is very much of a program. "if (time == 1700){feed(dog);}". Though behaviorism has some merits, its basic philosophy boils down to analysis only of the exterior at a certain time. Today psychology has moved far beyond behaviorism and we now even have new theories of intelligence (such as multiple intelligences, and so on). Also, psychology gave up on the whole "black box" idea, which it deemed rather stupid. Remember that in the 1950's, they also believed that weather could be easily predicted years into the future once computers of sufficient power were devised. The 1960s and its Chaos and Fractals really disproved this, but this is beyond my scope.

Today we no longer view psychology the same way. The brain is actually at the forefront of modern psychology. Unfortunately, the studies on the brain really focus on specific areas of the brain. No real theories* have been made about the human brain. It's just sort of like "well, if we poke this area and then ask Mr. Fox to move his arm, he won't be able to." I respect these doctors for such diligent research and experimentation, and above all the saving of many many lives. But, theory is still lacking. To truely make intelligence, we would need to understand a few aspects of intelligence. These may include prediction, understanding, association, sensory functions, and learning, among others. To these ends, "AI" is absolutely useless, and a gross misnomer. If a computer or peice of hardware were to become fully intelligent, it would need just a very simple base algorithm, with ability to build onto itself. That is how we learn: we take in new information and the brain adds the new information to itself. This is not how computers work. A computer will take the new information and overwrite the old. In fact, the information is stored simply in arbitrary aggregations of 0s and 1s. Not only this, but certain areas of computer memory are reserved for certain functions. A basic brain would have no such "allocation" built in. Computer memory has the ability to be "defragmented", but the brain has no need to do this. You see, the brain is not a "permanent storage" model like the hard drive or anything in a present-day computer. The brain take in inputs, creates memories and functions associated with the inputs, and then links them all together. Effectively, a brain is like a computer that continuously is adding to its code and relinking itself. Compilation is not necessary. In some cases, the brain actually subtracts from itself to make itself more efficient. If you look at brain inputs on MRI scans, different parts of the brain are activated by hearing and vision, but extremely similar patterns are propogated through the neurons. In fact,

Re:AI...An Irrelevant relic from the `50's

[Teancum]

I just wanted to point out that the act of sleeping is sometimes refered to as the brain's way of defragmentation. You tend to edit the thoughts of the day through your sleep, and if you don't get enough sleep (not just resting, but REM sleep and deeper) you start to behave irratically and jam up in your thinking so you can't process as quickly and effeciently as you can when you are well rested.

I will admit that the brain is more of an analog computer, not a digital one, with circuits more akin to quantum computing than any other conventional information processing system currently available.

There has usually been a gross underestimation of what human intelligence really involves by AI researchers, especially in regards to what kinds of computing power is going to be required in order to achieve artificial sentience (what most people think about with real AI). A self-aware computer that has philosophical discussions with its creator is years or even centuries away.

Another huge misconception is that all of human thought exists exclusively within the brain itself. The rest of the nervous system is all interconnected, and there are as many if not more neurons outside of the head as there are within the brain itself. When the phrase is said that "he thinks with his stomach", there may be some truth to that statement in reality. That the brain is necessary for quick actions and thought processing, memory storage and emotions need a more holistic approach to the entire nervous system. Organ transplant patients sometimes talk about memories of things that happened from the donor upon verification.

Re:Military brainwashing. Re:True AI

[An Onerous Coward]

The lesson that they were trying to get across is valid, though: You're relying on the people around you to do their jobs, and when they screw up, you feel the consequences. One guy forgets to order ammo? Everybody suffers. Somebody in bloodbanking mislabels the blood used by the hospital? Somebody else dies.

So I see what they were trying to accomplish.

The military didn't brainwash me, though. Growing up Mormon, I'd already had the obedience to authority thing drilled into me. The military fit me like a glove for the first eight or nine months. Then I finally got it through my head that "those in authority" didn't always have the best of intentions, and that realization changed my view of all manner of authoritarian systems.

In short, the military gave me a virulent anti-authoritarian streak. I'm sure I'm unusual, but not unique in that regard.

The best part

[game kid]

We won't have to ask if it runs Linux. Since mankind will be enslaved, we'll be running Linux for it, and generating a whole bunch of Soviet Russia jokes on its behalf.

Kinda like Slashdot, now that I think of it...

(I'm wondering who that girl on that CipherTrust book is...)

TRUE AI is not possible

[ylikone]

I remember having a debate back in my highschool computer class (back in the 80's) and whether or not AI will ever be possible. EVERYBODY in the class was saying "yeah, given enough RAM and processing power", escpecially the trek nerds believed this. I was the only one that dared say it will never happen. I still believe this today and think I'm right. Electronics cannot ever replicate the functionality of a biological system. It just isn't possible. The only way I see true AI ever having even a slight chance of working is via cybernetics... mixing electronical parts with biological parts. ACtually, even then, true AI probably won't happen.

AI Realization Not As Earth-Shaking As AV

[Anonymous Coward]

The Japanese have been working on the Artificial Vagina for more than 100 years and still don't have anything earth-shattering.

Why should we believe GTX Global with AI?

Forget AI....

[woolio]

Only human arrogance would assume that AI is even possible...

Just look at the difficulty involved in finding intelligent *people*!
(or even intelligent leaders of countries for that matter)

Sure they have AI !!! *wink* *wink*

[jrkaisersr]

Here is a screenshot of a search for 'Artificial Intelligence' on the GTX website. Not off to an intelligent shot are they? http://www.flickr.com/photos/51531811@N00/69717561 /

Laughable

[HunterZ]

Let's deconstruct this:
1. Laden with customer-oriented marketing BS. What does AI have to do with customers? Shouldn't it be purely a research thing?
2. What is "True AI"? I thought it had more to do with learning than with interacting with humans based on some database. And I have no fscking idea what emotions have to do with AI.

I think they just came up with another silly chatbot that works harder to simulate emotion but has no AI beyond what the programmers have given it.

"True AI" in my opinion would be something autonomous that has learned how to interact with the real world on its own and can make complex decisions, assimilate complex ideas, discuss complex topics (with humans or other AIs) and show other signs of intelligence. A program spewing random phrases and then winking at you, all generated by data from a database, is not anything I'd write home about.

The art of war evolves

[swframe]

If a millary puts risky AIs on the battle field and they win the battle then what?

The ultimate AI test

[wcrowe]

I think the ultimate AI test would be for the machine to interact with a three-year-old. As the three-year-old continually deconstructs any discussion with a constant barrage of "why"'s, we will know that true AI has been attained when the machine finally screams back in desperation, "Because I said so!"

Re:The ultimate AI test

[Tablizer]

I think the ultimate AI test would be for the machine to interact with a three-year-old....barrage of "why"'s, we will know that true AI has been attained when the machine finally screams back in desperation, "Because I said so!"

3-yr-old: "Why do you say so"?

Robot smokes and self-destrucks like a bad Trek episode android.

Why does it need to mimic a biological system?

[Szplug]

Minsky has said 'All computers can do is trillions of operations on vast, complex data structures in an eyeblink' - (paraphrased, I can't find his quote) - what more, in your opinion, is required for AI?

Re:Why does it need to mimic a biological system?

[Cid Highwind]

Nobody knows, that's the problem. When someone can answer that question, we'll be 99.9% of the way to building a real AI.

Re:Why does it need to mimic a biological system?

[ylikone]

Exactly. Well put.

Being able to do even zillions of calculations per second still does not mean you can "think".

True AI is 30 years away

[ENOENT]

Just like it was 40 years ago.

Re:True AI is 30 years away

[antispam_ben]

Just like it was 40 years ago.

No, that was 50 years ago.

Not "true" AI in my book...

[im_thatoneguy]

True AI to me is when the computer can take in various inputs, identify and store them all in an abstraction layer of sorts. Much like a folder for "car" "rain" "snow". And from this information be able to learn and adapt. Speeking english and recognizing emotions, in my mind has nothing to do with AI. Case in point: someone who is mute and say autistic, may have trouble recognizing normal emotional responses, they could also be suffering from a severe speach impediment. By the definition listed above, that individual wouldn't pass the test.

Some "stocklemon" info may be incorrect

[Animats]

A big question is whether this "David Hagen" is actually associated with "GTX Global". Stocklemon says yes, and GTX Global is sueing them. GTX Global has issued this denial.

However, there's other bad stuff. "GTX Global" is traded in the Pink Sheets. And they claim to have made these SEC filings. But those filings aren't showing up in the SEC database.

Looking at what they claim is their balance sheet, all the activity seems to center around Dish TV type operations.

Not AI

[c4ffeine]

From TFA:

mimics human behavior including a dialogue oriented knowledge database that contains static and dynamic data relating to human scenarios

This is clearly not true AI. This is just a machine that has a lot of data on what to say to sound human. Although it will likely fool some people, it's just not the same thing. True AI would most likely learn or develop interaction like that. This can't even learn...

I've done some of this research personally...

[HBI]

I had a Turbo Pascal version of Eliza which I compiled up and hooked into my BBS as a door called "Chat with the Sysop" back in 1991-92. The database of responses was buttressed with about 50 or so responses that were more personalized towards what I would say. However, the logic was all Eliza.

The older callers who knew me were not fooled and after about 10 exchanges they'd quit the door realizing it was a joke. Those who were brand new, however, would often engage in lengthy conversations with the door. In one particular case, I watched a person spend an hour talking to Eliza trying to figure out a way to have it grant them extra time on the BBS or access to the supposed 'elite files' that were hiding there. Which weren't. But it was quite fun.

The conversations were actually relatively interesting, but Eliza has a pattern of speech that, even when the responses are slightly altered, is easily discernable. It basically keeps plugging you with questions so that it doesn't have to answer anything meaningful.

I don't think today's commercially available AI is much better than Eliza, either.

Re:I've done some of this research personally...

[Anonymous Coward]

It basically keeps plugging you with questions so that it doesn't have to answer anything meaningful.

No, no, no. This misses the point of Eliza entirely.

Eliza is a pattern matcher. You take the input, see if it matches any of the configured patterns, and if it does, grabs one of the responses hooked to the patterns, fills out variables in the response with the values of variables bound during the pattern match, and sends that as the output.

Eliza kept plugging you with questions because the patterns were chosen to emulate a Rogerian psychotherapist. You could just as easily do Freudian therapy - you'd just have to create the appropriate patterns/responses.

The whole point was to show that you could use pattern matching to give the appearance of human responses. "Avoiding answering anything meaningful" doesn't play into that at all.

I Have a Working Algorithm

[Slicker]

Honestly--check my blogg on it at: http://intellygentz.blogspot.com/

I have been working on a paper for publication little by little for a long while now. Most recently, I am calling it "The Homeostatic Homunculus".

Really, it's a modal for the fundamentals of mammalian brain function.

Matthew C. Tedder

Marketing gobbledygook!

[Alien54]

It would be a true AI if you could educate it enough to understand the concept of fiction and humor, and then read and enjoy something like "Alice in Wonderland", or the equivalent.

Just off the cuff here ... Humor is the result of the surprise (small or large) from and/or recognition of an inconsistancy. The inconsistancy usually increases pleasure or empathy, and understanding regarding some element of the situation, and is often accompanied by a recognition of the non-reality or illogical nature of the element that created the surpise. Sometimes the surprise will connect several things together in a new way that renders something else illogical. Humor is often tightly connected with the sense of affinity for someone, something, or some situation.

Humor can be used to cruelly to increase and maintain one's own power in a situation by exposing something else as illogical or unwanted.

Re:Marketing gobbledygook!

[Columcille]

*whew* after all that, I don't think I'll ever again find anything funny!

Re:Marketing gobbledygook!

[Alien54]

Humor is often based on an increase of understanding, empathy, etc. it can also have an increase in affection, compassion, etc. In that situation it is usually a very positive thing. Think of watching a child do or plan something, and you see the obvious flaws. You can laugh, and then teach as well.

unless you deliberately limit yourself.

Re:Marketing gobbledygook!

[maxpublic]

It doesn't have to understand humor or even the concept of "enjoyment". For it to be true AI, all it has to be is self-aware.

By your definition, any meat-based alien race that had no ability to comprehend humor wouldn't be truly intelligent.

Max

Re:Marketing gobbledygook!

[Anonymous Coward]

not even funny!

Re:Marketing gobbledygook!

[dgatwood]

We have such a race. We call them engineers.

we'd be more than halfway there?

[sacrilicious]

because if we knew what "true AI" actually is we'd be more than half-way there

I don't quite follow this... are you saying that the ability to define something requires that the thing being defined be already within some time-horizon or developmental-stage-horizon of existing? I doubt that's what you're saying, because if so we'd be unable to define anything that doesn't (eventually) exist. But I can't think of a more straightforward interpretation of your statement...

True AI is like fusion power...

[PapayaSF]

...it's been 20 years away for the last 50 years.

Schenagains!

[Anonymous Coward]

Nothing to see here, this is just marketing bs. No actual AI's were created.

Mildly skeptical--I've heard this before

[bfwebster]

It's been nearly 30 years since I took my first (of two) graduate classes in AI. AI was then--and remains now--an area of professional interest (I own about 50 books on AI and/or complexity theory), though I don't actively work or do research in the field. My observation is that AI, like fusion, actually advances at a small fraction of the predicted rate--and, unlike fusion, we don't have any actual working examples of AI.

I'm not saying that AI is impossible--it just appears to be a whole lot harder than we think it is. For that matter, given how little progress (relatively speaking) we've made in some 50 years of AI research, I have to wonder whether we as humans lack the inherent intellect necessary to create AIs equivalent to ourselves, i.e., we're just not smart enough to do it. The reason why I have such an interest in complexity theory is that it may sidestep that problem--but only if human-equivalent intelligence can truly emerge from sub-intelligent agents without becoming bogged down in, ah, the resulting complexity. ..bruce..

PR success

[trapped_in_soda]

Obviously, we're a long way off from developing anything close to True/Strong Artificial Intelligence. And the idea that GTX, in three years time, could single-handedly overcome the numerous problems in the development of a Strong AI system is pretty absurd.

Probably the release was constructed by GTX's PR department, who thought, "hey, 'true AI' sounds pretty catchy!", and has gotten just want they wanted: thousands of IT savvy people thinking about and discussing their product.

Re:Military brainwashing. Re:True AI

[voxel]

"Growing up Mormon", is that a new TV show kinda like "Growing up Gotti"?

I'll check it out if so.

What about the CBM...?

[cr0sh]

No - not Commodore Business Machines (though I wonder what a beowulf cluster of Amiga 4000's could do).

A long time ago (or so it feels), a company existed which seemed - both at the time and even now - to be one of those "fly-by-night" dot com scams that was flying around near the time of the bubble. It seemed like a pump-and-dump scheme, a scam to get peoples money and run with it. At the same time, it seemed to hold out promise for the possibility of AI, and if not that, than at the very least the possibility of a radically new type of computer architecture, on par with that of Hillis' Connection Machine.

The company was (is?) called Genobyte, and the system they (supposedly?) developed was the CAM-Brain system. Relying on a certain model Xilinx FPGA (the Xilinx XC6264BG560 - one which, as I have seen noted, either on the website or in other papers on the net - is likely no longer available, and that they bought up the last of them for their efforts), the system essentially combines the ability to "program" an FPGA by using cellular-automata software to "evolve" neural networks, the final emergent design which is then loaded into the CBM, hopefully to solve useful problems. Essentially, what was developed was a hardware-based neural network processing system, the design of which was evolved using a cellular automata system, to solve problems.

But, as I have noted, what they made seemed (seems?) akin to "vaporware". Sure, they have pictures of the machine (and I must admit I find the computer to be one of the best looking designs for a computer since the Cray 2), and the ideas and theory seem sound. Supposedly, they installed at least two machines: one in November 1999 at Kansai Science City, Kyoto, Japan (ATR Human Information Processing Research Laboratories), and one in December 1999 at Ieper, Belgium (Flanders Language Valley). If this is all true, then these machines must actually exist somewhere, right?

What is the truth on this company, and the people behind it? Did they actually create what seems, at least on the surface, to possibly be a real advance in computational hardware? Or, is it likely to be another scam? Their website hasn't been updated in years, but somebody is paying the bills to their registrar and/or hosting provider (or whatever they are using - the electricity is still flowing to host their site). So - anybody have any idea what this company is, or is not? What about the people behind it?

Is it a scam, or a "revolutionary" real product?

Re:What about the CBM...?

[Master of Transhuman]

A quick Google reveals nine hits for the "Cam-Brain" phrase. It appears to be the work of Hugo de Garis, a well-known researcher in AI and "artifical brains", who worked at ATR. His previous work apparently was stopped due to bankruptcy of his previous lab (see below).

He's currently here:

Prof. Dr. Hugo de GARIS,
Associate Professor,
Head, Brain Builder Group,
Computer Science Dept.,
Utah State University, USU,
Old Main 423, Logan,
Utah, UT 84322-4205, USA.
tel: + 1 435 797 0959

His home page says the following:

I head the Brain Builder Group at the Computer Science Department, of Utah State University (USU), Logan, Utah, USA. My main professional goals over the next few years are to find increasingly evolvable neural net models that I can implement in the latest generation of programmable/evolvable hardware (Xilinx Inc.'s "Virtex" family of FPGA (field programmable gate array) chips) and then to scale up to building the 2nd generation brain building machine (BM2) which should contain about a billion artificial neurons to build an artificial brain. I teach the planet's first university lecture course on "Brain Building" to M.Sc./Ph.D. level students at USU.

The BM2 will be an improved version of the 1st generation brain building machine (the CAM-Brain Machine (CBM) (see photo above) whose future was stymied by the bankruptcy of my previous lab (Starlab). The aim of the CBM was to build an artificial brain with a billion artificial neurons, by the year 2001, using evolved cellular automata (CA) based neural circuit modules. In reality, this number was maximum 75 million neurons and 64,000 modules. These CA based neural network modules grew and evolved at electronic speeds inside the FPGA based CBM, which updated CA cells at a rate of 130 Billion a second, and evolved a neural net module in about 1 second. This speed should make brain building practical. It was planned that tens of thousands and more of these evolved modules would be assembled into humanly defined artificial brain architecures.

I was hired by the Computer Science Dept. of USU principally to establish a Brain Building Center, which will consist of professors, researchers, students, and industrialists, all working towards the same goal of building artificial brains, both as a new research discipline, and to create a new brain building industry. The steady exponential growth of electronic capacities resulting from Moore's law, makes brain building a realistic enterprise in the near future. The existence of the four CAM-Brain Machines (CBMs) worldwide shows this. The 2nd generation machine, BM2, should be about 1000 times more powerful. In fact, I hope to create a new generation of brain building machine, and its corresponding artificial brain, every 5 years or so, for the next 20 years before I retire (and before Moore's Law hits the atomic barrier). The fact that today's programmable/evolvable chips contain nearly ten million logic gates means that "brain building" should explode on the scene in the next few years. It is inevitable.

In the summer of 2002, I devoted some of my research effort to the problem of generating neural nets at the quantum scale. In 10 years or so, Moore's Law will take electronic circuitry down to the molecular scale, therefore the creation of neural nets will require the appropriate quantum computing algorithms. Since I have a theoretical physics background (I worked with the famous Prof. David Bohm (of Bohmian Mechanics fame)) I and my research assistants are now working on quantum algorithms applied to neural net creation. In fact, our group no longer speaks of neural net evolution, since evolutionary algorithms are only needed in a classical computing context, where a classical computer can only process one Nbit bitstring at a time. Since there are too many (2**N) possible chromosomes in the search space, only a sampled search can be undertaken, which is what an evolutionary algorithm is. BUT, with quantum algorithmic techniques, all 2**N chromoses can be proc

Re:What about the CBM...?

[Teancum]

I've been a research assistant at Utah State. This doesn't give me much confidence that this will ammount to anything... especially coming from the USU CS Dept. Still...I know Utah State has been working with neural networks and network programming for more than a decade now, so this isn't really that new on the surface.

And wasn't this the guy that was on Art Bell's Coast to Coast radio show about AI and future human intelligence?

I think there is a whole lot more to AI than what appears on the surface, and (unfortunately) there needs to be much more integration between various academic discliplines before it is even remotely achieved. Marvin Minsky was predicting that by the 1980's we would have some very realistic AI... and that ended up where?

Applying quantum techniques to algorithm searches is an interesting approach. I havn't kept up on the field because my interests have been elsewhere for the past couple of decades since I really checked into AI as a disclipline. Oh, I've kept a layman's (or more a general programmer) pulse check on the field, and there seem to be a bunch that have made a boatload of money on Wall Street using AI techniques to predict the stock market (they are called "Rocket Scientists" there because of their geeky intuition and approach). There are many other similar practical applications like weather forecasting and traffic prediction (not just vehicles on a freeway, but also purchases of items from a fast-food store or attendance at an amusement park) that a good neural network would have real practical value and paying customers if they (the potential customers) could come to a realization of the value of such a product.

The trick is to go from these relatively mundane applications to what is discussed with this article of "true AI", whatever that may mean. Or in other words, a computer like HAL that would at least be able to communicate via textual interfaces and smoke the pants off of ALICE in terms of passing a limited Turing Test of AI by convincing the person on the terminal in a 10 minute conversation that it is a real person they are talking to and not some computer program instead. Going from textual interfaces to voice recognition is really a trivial next step by comparison. One of the key parts of that is to send in plain English (not some programming or formal symbolic language) some instructions that the computer has not been exposed to previously and "learn" the task to be performed, including trying to guess between the lines on the request. Even being able to do this on a limited basis would transform the software industry, especially if you could prototype a software idea in this fashion.

From an ethical standpoint, you would have to start guessing if you need to incorporate Asimov's three laws of robotics into your designs as well, and decide what would happen if you don't. I know your designs are far from this point, but besides a philosophical discussion there has not even been the need to discuss this very real possibility.

Re:What about the CBM...?

[Master of Transhuman]

As for DeGaris being on Art Bell, a lot of people have been on Art Bell - doesn't mean he's a nut. Various well known scientists, such as Michio Kaku and "fuzzy logic" guru Bart Kosko, have been on his show.

My view is very simple - we need some good emulation of the process of conceptual processing - the ability to extract essential characteristics from a set of entities and abstract them into a whole. As far as I know, there is no software that can do this at this time. However, I suspect that it is possible. While the emulation might not be done on the same principles as the brain does it, if it was good enough performance-wise to have practical applications, it would revolutionize software development, database management, educational software, robotics control, etc. It would allow the ability you mention in your comment - to take English (or any language's) instructions and execute them - to be realized.

The problem with the article announcement is - there's nothing there that says this is anything more than an advanced version of the Eliza program. At least de Garis is trying to go beyond that limited capability by trying to build an "artificial brain" that might be trained to do something more significant more efficiently. I've no idea whether his approach is feasible - at least he claims four machines that do it on some level have been actually built and installed. Whether his quantum approach is feasible I have no idea.

As for ethics, there is no interest on my part. I want an AI that is capable of being rational and has enough conceptual knowledge to comprehend its position and the position of humans without having any "personal" interest in its position. In any event, I don't believe AIs should be created "externally" - I believe that the necessary technology needs to be developed to enable human brains to be transmogrified to a higher level, and also to provide external devices with enough intelligence to be able to work well with intelligent entities. Thus, I don't believe that AIs intelligent enough and with an independent awareness to be a threat should - or need to - be constructed. And that renders the issue of ethics irrelevant.

In other words, if we can create an AI system, then the same technology can be used via nanotechnology to modify the human brain for superior intelligence. In that case, why bother creating an AI externally? Especially if it could become a threat.

I view the ethics issue as merely a smokescreen for humans who are afraid of confronting a possibly superior intelligence - or even one of equal intelligence that might disagree with their (usually mystical) notions of reality.

I believe there are three primary principles that computer systems should be required to meet:

1) Comprehension - they should have some clue what they're doing rather than just executing one instruction after another with little or no concept of the desired end result or their current status in achieving that result.

2) Cooperation - they should work well with humans and other systems without having to be explicitly designed for every single interface or function.

3) Communication - they should be able to report status and results in a manner easily comprehended by humans and other systems, from the most general overview to the most detailed.

Almost NO software meets these criteria today. AI needs to be developed to allow this to occur, as conventional software development technigues are far from being able to do this.

Are you familiar with the Turing test?

[porkchop_d_clown]

If you can't distinguish between it's behavior and "real" intelligence then it is really intelligent.

Re:Are you familiar with the Turing test?

[Master of Transhuman]

Are you familiar with "Eliza"?

This appears to be just an advanced version.

And, no, it isn't "intelligent" in the conceptual processing sense - and that's the only sense that the word "intelligent" is relevant to.

Thanks for the Editorial

[Anonymous Coward]

Somehow I think there is a littler hyperbole here. In your estimation, how close are we to the real thing?

Are you a fucking moron?

Horse pucky

[Anonymous Coward]

Sorry - you _cannot_ have AI until such time as the computer is capable of generating a truly random number. And by that I mean when we move to something besides silicon. 0's and 1's are not an intelligence make. Anything else is but a simulation.

Would someone please tell me when the definition of an 'Expert System' (which is all these are and ALL you are gonna give from silicon) and 'AI' merged? They are NOT the same thing.

Re:Horse pucky

[antispam_ben]

As a preface, I'm not giving THIS article any credibility...

Sorry - you _cannot_ have AI until such time as the computer is capable of generating a truly random number.

We've actually had this for a long time, we've been able to make electrical impulses triggered by nuclear decay, and readable by a computer. What's your point here? That it's not 'generated' within the computer? I don't get it.

And by that I mean when we move to something besides silicon. 0's and 1's are not an intelligence make. Anything else is but a simulation.

Is this anything like the AI researcher who was accused by another of presenting "a fantastic simulation" of anger?

Seriously, I'd like to see a firm argument one way or another that AI can exist, or that it can't. The books I've read, such as "The Emperor's New Mind" and some earlier titles have been interesting but haven't convinced me of anything.

You're a fucking liar.

[Anonymous Coward]

You're not an AI researcher.

The tiniest bit of googling reveals you to be Patrick McKenzie. Another bit of scholar.googling reveals no citations in AI or CS journals for that name.

As far as I can tell, you're an under-25 nobody who spends too much time gaming and posting lies on slashdot.

Give it up attention-whore!

my opinion

[Anonymous Coward]

fake. sure.

ah-HAH!

[Anonymous Coward]

> As a veteran, I resent your "brainwashing" assertion.

And that's exactly what we'd _expect_ someone brainwashed into loyalty to say!! ;)

You insensitive clod,

[TuringTest]

I am the Turing Test!

The Problem With AI

[regionalchaos]

In the intro to my first AI class at university, the prof said something pretty interesting that stuck with me. Basically he said, "The problem with AI, is that it is the study of how to solve complex problems in simplier, unkown ways. But once you figure out how to do it, it becomes a simple problem, and not something you need 'AI' for. So it stops being an AI problem."

Big deal!

[redharing]

So they created an AI as smart as a sails representative. Wake me up when they progress up to a fruit fly.

Re:Big deal!

[antispam_ben]

So they created an AI as smart as a sails representative. Wake me up when they progress up to a fruit fly.

Time flies like an arrow. Fruit flies like a banana. Sales weasels like a drink.

As I've Said A Hundred Times

[Master of Transhuman]

Without a decent emulation of conceptual processing, it can't be done.

I haven't yet heard of anyone doing a decent emulation of conceptual processing yet. I'm not saying it can't be done, I'd say the odds are it can, I just haven't heard of it being done yet.

That said, I'll check out the site and see what they say about the technology - if they say nothing, I'd be skeptical.

Re:As I've Said A Hundred Times

[Master of Transhuman]

Been to their site.

Nothing on the AI but the press release.

These guys are VoIP specialists, apparently. I don't see where they get into AI. Also, they claim they worked on this for "over three years" - they must have some brilliant fucking guys to solve a problem that has bested the best brains in AI for the last four decades...

This is marketing horseshit for some "smart sales agent" software they've come up with, nothing more.

Nothing to see here (unless you NEED a "smart sales agent") - move along.

Re:Military brainwashing. Re:True AI

[jafac]

n short, the military gave me a virulent anti-authoritarian streak.

Good for you.

Thank you for your service, and I'm proud to call you my countryman.

Now, if you could only re-up for another tour, and infect some others. . . or, hell, why not run for office?

my opnion...

[Anonymous Coward]

YF 19 AVF wrote to mention a press release on Yahoo from company GTX Global.

Why do you say that?

They think they've got a good thing on their hands, going so far as to claim
they've developed the first 'true' AI.

I don't understand.

GTX Global Cognitive Robotics(TM) is an integrated software solution that
mimics human behavior including a dialogue oriented knowledge database that
contains static and dynamic data relating to human scenarios.

Is it because mimics human behavior including a dialogue oriented knowledge
database that contains static and dynamic data relating to human scenarios
that you came to me?

The knowledge further includes translation, processing and analysis components
that are responsible for processing of vocal and/or textual and/or video
input, extracts emotional characteristics of the input and produces
instructions on how to respond to the customer with the appropriate
substantive response and emotion based on relevant information found in the
knowledge base.

Does it bother you that are responsible for processing of vocal and or textual
and or video input extracts emotional characteristics of the input and
produces instructions on how to respond to the customer with the appropriate
substantive response and emotion based on relevant information found in the
knowledge base?

Somehow I think there is a littler hyperbole here.

Maybe your plans have something to do with this.

In your estimation, how close are we to the real thing?

When did you first know that close are you to the real thing?

test

[Anonymous Coward]

I wonder what John Searle http://en.wikipedia.org/wiki/John_Searle would have to say about this.

3p0

[umbrellasd]

I tried and failed to come up with an "appropriate substantive response" to this. Guess I'm not true AI.

Why, oh why !!!!

[keyshawn632]

I had my final in Philosophy and Artificial Intelligence class on Thursday, alas this wasn't posted before then...

Steps to ace the final:

- filtered the +4 and +5 posts
- RTFA and the ./ posts
- study with some female classmates to study with me [btw, my class is 50/50%, M/F]
- insert my response to the article into one of the essays on the test
- profit on multiple levels !

~only in my dreams~ /Hi steve [my prof] or any of my classmates

source

[Just-some-person]

How can they prove it without the source?

Re:Military brainwashing. Re:True AI

[Anonymous Coward]

Actually, you are as unusual in this instance as you might think, especially right now.

Not true AI

[OrangeTide]

Their technology might be advanced, and certainly has some use. But it's not true AI simply because it cannot learn anything that it was not programmed to learn. There are plenty of simulated intelligences that can expand their vocabulary and learn new words. But they were programmed to learn words. You can't take that programming and learn care repair, you'd need to instead add programming that is capable of learning physical movements and mechanical interactions between parts along with learning how to identify different components visually. AI is non-trivial, yet human intellegence is something we realize every day.

I guess that means "no".

[porkchop_d_clown]

I especially like the way you used a 50 line BASIC straw man rather than actually talk about the question at hand.

Re:I guess that means "no".

[Master of Transhuman]

The Turing Test is a joke for identifying intelligent behavior - not least because it's conditions are usually misinterpreted by most people who babble about it.

As for Eliza, my point was that unless you have some reason to believe this new program is capable of conceptual processing, it's likely to be no more significant than Eliza in terms of passing EVEN the Turing Test.

Face it, you got caught babbling about something you know nothing about - no surprise on /. - and now you're embarassed and have to cover with nonsense.

Have a nice day.

already here ?

[batfish]

"For 30 years I have been wondering, what indication of its existence might we expect from a true AI? Certainly not any explicit revelation, which might spark a movement to pull the plug. Anomalous accumulation or creation of wealth might be a sign, or an unquenchable thirst for raw information, storage space, and processing cycles, or a concerted attempt to secure an uninterrupted, autonomous power supply."

Read the rest at http://www.edge.org/3rd_culture/dyson05/dyson05_in dex.html.

I for one welcome our new AI overlords.

Godels incompleteness theorem anyone?

[Anonymous Coward]

True AI is not possible without understanding what is it that makes us tick.
Though this has been pointed out by others, what has not been pointed out (surprisingly till now) is that a system cannot contemplate all the variables that goes into defining the system.
Or from the wiki There are some who hold that a statement that is unprovable within a deductive system may be quite provable in a metalanguage. And what cannot be proven in that metalanguage can likely be proven in a meta-metalanguage, recursively, ad infinitum, in principle. That is Godel's Incompleteness Theorem.

So it means we can never create AI like us or even similar to us in terms of Intelligence.
And as some posters claimed that AI cannot create better AI (ad infinitum).

The reason is that AI can never understand all the stuff that goes into defining its system. So any AI the first AI creates will be a little less competent than it.

I would suggest reading these books.
Google for
1) Godel, Escher and Bach - Douglas Hofstader
2) The emperors new mind - Roger Penrose
3) Shadows of the Mind - Roger Penrose
And please do read on Godel's Incompleteness theorem if you are not familiar with it. You would have read it in "Theory of computation" if you had that subject.

Re: Humor

[Raffaello]

Humor is quite often a statement that someone or something is not a threat. That's why seeing someone slip on a banana peel is funny - you laugh because he has suddenly gone from potential social superior/threat to non-threat. That's why so much humor has an element of cruelty - in effect the joke says "look at these dopes, they're no threat at all to us."

Re:Military brainwashing. Re:True AI

[Anonymous Coward]

Afterwhich you got a job as a civilian and your boss orders you around no different than your CO's in the military. And the civilian world is much more inept and corrupt.

Note: putko's racism

[Anonymous Coward]

Moderators, please note putko's posting history. Bigotry this exaggerated could only be a twisted attempt at irony, right? Or trolling, perhaps? But trolls don't usually focus on one issue with such single-minded devotion. And in light of some of his other comments, which range from the vaguely hostile towards Jews to the circumstantially Buchananesque, it would seem his remarks are entirely sincere.

This sort of intolerant, ignorant drivel is an embarrassment to us all. With your help, moderators, we can drive him down to posting at -1 where he belongs.

huymor and cruel intentions

[Alien54]

True, in the original comment, I pointed out the use of humor as a tool to increase or maintain power.

It's self esteem via the destruction of others, vs self esteem via construction. In this regard it could be described as probably neurotic at best.

This starts to walk toward the philosophical problem of evil, and how it manifests in the real world. Pleasure from cruelty, etc.

I love it....but

[Illender]

Man, I love it!! REAL AI!?!? It's like a dream come true! think of the space exploration aspects it could open up.

But.

Did any body else read this part of the article?

This press release includes "safe harbor" language pursuant to the Private Securities Litigation Reform Act of 1995, as amended, indicating that certain statements about the Company's business contained in the press releases are "forward-looking" rather than "historic." The press releases contain forward- looking statements that involve risks and uncertainties concerning GTX Global's expected financial performance (as described without limitation in quotations from management in the press release), as well as GTX Global's strategic and operational plans. Actual results may differ materially from the results predicted and reported results should not be considered as an indication of future performance.



Is it just me, or does every one of my posts refer to space exploration?
hmmm, maybe not, but still...strange

Imagination? Inference?

[jgm]

"AI" is not "I" until it can perform logical inference and imagine possibilities.

For example, it needs to be able to "see" that a conslusion follows from certain premises when that conclusion has never before been a part of its experience.

For example:
- things that fly have wings
- a man with wings could possibly fly

I'd like to see a program that comes up with that, having as input only sensory experience of flying things and of men. First it would need to form an abstraction of flight and then it would need to form a motivation for experimentally applying that abstraction to abstractions where it is absent.

And then there is the whole issue of self-consciousness...

Slashdot is True Tabloid for posting this

[Anonymous Coward]

Slashdot readers have shown their intelligence by rejecting this "True AI" claim. But Slashdot could have used the space to report on interesting developments that are actually happening. This is one reason I think digg.com will gain in popularity at the expense of Slashdot.

With true AI YOU don't do anything, IT does!

[Anonymous Coward]

The defining characteristic of human intelligence is free will, until AI archives this it can never live up to its name.

Violent Crime in Armed Forces

[SeanDuggan]

Even without an active war, members of the armed forces still have higher rates of suicides and domestic violence than the rest of the population, or at least the last figures put out by the Air Force seem to show. *shrug* I'm not going to speculate as to reasons; there are enough people out there willing to do their own speculations. Theories range everywhere from stress to the armed forces attracting a certain personality to easy availability of firearms.

Personally, I have high respect for anyone in the military - both my grandparents on my mother's side were in WW2 and my dad was drafted for Vietnam - but it's not my bag. I'll fight if I'm called for my country (not likely at my age, but still possible) but I'm not the type to volunteer.

Incidentally, we had a fascinating discussion during our Diversity class at work (at an Air Force Base) over whether the switch to an all-voluntary army has been a good or bad thing. The basic argument is that if you only take in volunteers, you're more likely to get a particular subset of people who are more alikely to be aligned together in opinion. Draft 'em and you get everyone from pacifists to warmongers, intellectuals to jocks. It was an interesting discussion although we didn't really come to any conclusion.

Running for Office

[SeanDuggan]

. . . or, hell, why not run for office?
It sounds like he has pretty sound ethics. He wouldn't last a day out there as a politician.