You visit the Bonneville Salt Flats, install a racing game, keep the car in low gear so you don't actually roll it, and then boom! Your iPad racing game now has force feedback.
Claiming that it's not even a viable product launch because it lacks a solid pre-existing market share is a pretty broken argument. Calling the Zune a "potentially dying platform" is just baiting, because ANY platform is "potentially dying" to the point that at least SOME degree of investment could be considered unjustifiable.
You're just throwing out a bunch of speculative claims and poorly-supported attacks under the guise of combating bias, which is deeply, deeply ironic.
You've hit the nail on the head. I'm actually surprised that the Pentagon came out and said this instead of some hack writer for a tech rag. It's simply a matter of requiring security clearance, and there are PILES of contractors all over the country that require Secret or higher clearance just to even WORK there, and that basically means you need to be a citizen and you need to go through a process of background checking that can take up to two YEARS. If you failed to account for a single day of residence in the past 10 years, they'll catch it. They'll also verify every last residence with references, check your family and friends, interview acquaintances, and generally go over everything you've done in the past decade with a fine-toothed comb. Not a lot of room in there for an Al-Qaeda cyberterrorism training seminar.
I guarantee that if such an interface existed, no one would last a day without turning off the blur. Sometimes you read off a window that isn't necessarily in focus.
Re:Basic AI research important
on
Infinite Games?
·
· Score: 1
Actually, you've just redefined AI to be an easier problem. AI can't work yet, so you've picked some related research and "promoted" it to be called AI.
It seems to me, no offense, that you see AI as nothing more than the simulation of a human at a very high level. Current research is so far away from anything even close to this that the field of AI is, at least in the educational form, surprisingly little more than a set of widely accepted algorithms. Bayesian belief networks, A* searches, and the like. Research AI is in many cases simply the application of these established (and special-tailored) algorithms towards novel goals in the general direction of the approximation of either specifically human or otherwise simply rational reasoning. An "AI" algorithm by your definition, which could perform well on an SAT or similar aptitude test, is in some cases a trivial task compared to the over-arching goal of developing some ad-hoc reasoning within an agent. "Intelligence" is a far broader concept than simply "acting like a sentient being." Without getting into philosophy, the field of AI is divided into those who believe that the goal is to emulate human behavior and those who believe that the goal is to exhibit rational behavior. Since both of these viewpoints exist within the pre-existing field of artificial intelligence and that the argument is one of goals rather than foundations, it seems to me that we cannot exclude either interpretation as "wrong" but must instead see them as different approaches, and therefore both equally valid. Your argument is exclusively human, which effectively states that a huge amount of AI research is nothing more than algorithms research that just happens to take into consideration the human aspects of rationality.
"Articifical" "Intelligence". Anything else is incorrect. The expansive definition may be more useful for you in your research, but it is linguistically wrong. Your examples only use parts of a potential future AI system.
This pheomenon is fairly common, and a little interesting. A new thing is invented, with a new name, and a professional class springs up to study and improve it. As they work, they decide to focus on something other than what the word originally meant, so they create a new definition for their own use.
One can argue linguistics and semantics endlessly. I believe that my summary--that AI is an attempt to approximate (human|rational) reasoning--holds up with respect to the accepted definition of the term as a whole (i.e. in the context of computer science) and not as the literal interpretation of "artificial" and "intelligence" separately. In my personal opinion, AI isn't necessarily intended to approximate human reasoning as much as it is intended to approximate any "logical" reasoning. While simulating human reasoning may win the Turing test, it does little good to fool a remote typist into thinking your computer is a person. Meanwhile, decisions made by AI algorithms about gene sequencing may follow guidelines for rational decisions that are of a much higher-order of complexity than a human could ever comprehend. But all applications of AI principles in the context of making logically sound decisions fall under the category of artificial intelligence.
No, there are many more reasons to make a computer act like a human. Economic predictions, military simulations...
I'll admit that that statement of mine was flawed, but not so drastically. I meant something more along the lines of simulating an actual human being's behavior, rather than "acting like a human," which falls under the whole "rational decisions" umbrella I discussed previously. Under this revision, my statement still does not work completely as is, and in fact falls far short of categorizing all applications of behavioral simulation as such. But my point is that AI reaches far beyond simply simulating the actions a human would make. Natural language processing is great for HCI purposes and so on, but genetic algorithms are an AI concept and are used within distinctly non-human contexts more often than not.
Re:Slashdot is aiming way too low
on
Infinite Games?
·
· Score: 1
This article is a thumbsucker for technical illiterates, and an insult to Slashdot's readership.
Just because an article isn't techie-savvy doesn't mean the content is unworthy of consideration. The project is in fact an interesting and novel one, and the article at least brings the efforts of a relatively obscure project to the public.
My AI faq [robotwisdom.com] gives infinitely more perspective than this [short, introductory, made-for-the-masses] BBC pap, on the important questions.
I looked at your page and found four things: 1) Vast quantities of ego masturbation 2) Same stuff I can find in any AI book worth a damn 3) Same stuff I can find in 20 minutes on Google 4) Even vaster quantities of ego masturbation
Pardon me if I'm unimpressed.
Re:Answering my own questions
on
Infinite Games?
·
· Score: 1
It is. I was [sorta] one of those AI guys.
Re:Basic AI research important
on
Infinite Games?
·
· Score: 1
Actually, you're wrong. I'm sitting here in the Knowledge Discovery Lab at NCSU, and I can tell you that AI is almost ubiquitous in the high-tech industry as a whole. It is used in incredbly complex data visualizations, like those researched by Dr. Healey in the next room from me. It's used in network security, as in several projects by Dr. Doyle down the hall. It's even used in molecular biology, like Dr. Bahler's work. AI isn't just about making things act like humans. That's only good for human-computer interaction (Dr. St. Amant, my advisor, also just down the hall) and entertainment (Dr. Young, two buildings down). AI is about allowing computers to use concepts approximating human reasoning to solve problems, and those problems range from telling stories to determining how proteins fold.
Re:For the geeks...
on
Infinite Games?
·
· Score: 3, Informative
Actually, I was in Mimesis for a semester, and it's none of that, really. The only basis that I'm aware of was that UT can be modded heavily without spending thousands on licensing the engine. They've got a LISP server they connect to that manages all the game logic and the client is actually mostly a bunch of basic mod stuff, like tourists in the aquarium that wander around and form groups (which I coded) and various HUD changes and so on. All the real AI is developed on the remote controller, so it doesn't require direct modification of the UT engine.
Donkey Kong games aren't a dime a dozen? A Bioware-developed SW game is hardly something to ignore. Fable is hardly vaporware, as screens and trailers are everywhere now. You cut down games because you don't know what they are, yet cite a series of games you've never even seen, much less played. Have fun with all of your sequels of sequels of sequels that you've never even played yet but will OBVIOUSLY be so much better and more original than every Xbox game out there.
I'm currently enrolled at NCSU, so I may be able to clear things up a bit... I'm not entirely sure, but I think it was a professor (Dave McAllister, who taught my Graphics course a couple semesters ago) who was doing research with a company called LIPSinc (http://www.lipsinc.com). Their demos show promise, but it's still far from realistic. Hope this helped.
PHoRD
Slashdot Mirror
You visit the Bonneville Salt Flats, install a racing game, keep the car in low gear so you don't actually roll it, and then boom! Your iPad racing game now has force feedback.
Claiming that it's not even a viable product launch because it lacks a solid pre-existing market share is a pretty broken argument. Calling the Zune a "potentially dying platform" is just baiting, because ANY platform is "potentially dying" to the point that at least SOME degree of investment could be considered unjustifiable.
You're just throwing out a bunch of speculative claims and poorly-supported attacks under the guise of combating bias, which is deeply, deeply ironic.
You've hit the nail on the head. I'm actually surprised that the Pentagon came out and said this instead of some hack writer for a tech rag. It's simply a matter of requiring security clearance, and there are PILES of contractors all over the country that require Secret or higher clearance just to even WORK there, and that basically means you need to be a citizen and you need to go through a process of background checking that can take up to two YEARS. If you failed to account for a single day of residence in the past 10 years, they'll catch it. They'll also verify every last residence with references, check your family and friends, interview acquaintances, and generally go over everything you've done in the past decade with a fine-toothed comb. Not a lot of room in there for an Al-Qaeda cyberterrorism training seminar.
I guarantee that if such an interface existed, no one would last a day without turning off the blur. Sometimes you read off a window that isn't necessarily in focus.
juuri doody?
like this?
Actually, you've just redefined AI to be an easier problem. AI can't work yet, so you've picked some related research and "promoted" it to be called AI.
It seems to me, no offense, that you see AI as nothing more than the simulation of a human at a very high level. Current research is so far away from anything even close to this that the field of AI is, at least in the educational form, surprisingly little more than a set of widely accepted algorithms. Bayesian belief networks, A* searches, and the like. Research AI is in many cases simply the application of these established (and special-tailored) algorithms towards novel goals in the general direction of the approximation of either specifically human or otherwise simply rational reasoning. An "AI" algorithm by your definition, which could perform well on an SAT or similar aptitude test, is in some cases a trivial task compared to the over-arching goal of developing some ad-hoc reasoning within an agent. "Intelligence" is a far broader concept than simply "acting like a sentient being." Without getting into philosophy, the field of AI is divided into those who believe that the goal is to emulate human behavior and those who believe that the goal is to exhibit rational behavior. Since both of these viewpoints exist within the pre-existing field of artificial intelligence and that the argument is one of goals rather than foundations, it seems to me that we cannot exclude either interpretation as "wrong" but must instead see them as different approaches, and therefore both equally valid. Your argument is exclusively human, which effectively states that a huge amount of AI research is nothing more than algorithms research that just happens to take into consideration the human aspects of rationality.
"Articifical" "Intelligence". Anything else is incorrect. The expansive definition may be more useful for you in your research, but it is linguistically wrong. Your examples only use parts of a potential future AI system.
This pheomenon is fairly common, and a little interesting. A new thing is invented, with a new name, and a professional class springs up to study and improve it. As they work, they decide to focus on something other than what the word originally meant, so they create a new definition for their own use.
One can argue linguistics and semantics endlessly. I believe that my summary--that AI is an attempt to approximate (human|rational) reasoning--holds up with respect to the accepted definition of the term as a whole (i.e. in the context of computer science) and not as the literal interpretation of "artificial" and "intelligence" separately. In my personal opinion, AI isn't necessarily intended to approximate human reasoning as much as it is intended to approximate any "logical" reasoning. While simulating human reasoning may win the Turing test, it does little good to fool a remote typist into thinking your computer is a person. Meanwhile, decisions made by AI algorithms about gene sequencing may follow guidelines for rational decisions that are of a much higher-order of complexity than a human could ever comprehend. But all applications of AI principles in the context of making logically sound decisions fall under the category of artificial intelligence.
No, there are many more reasons to make a computer act like a human. Economic predictions, military simulations...
I'll admit that that statement of mine was flawed, but not so drastically. I meant something more along the lines of simulating an actual human being's behavior, rather than "acting like a human," which falls under the whole "rational decisions" umbrella I discussed previously. Under this revision, my statement still does not work completely as is, and in fact falls far short of categorizing all applications of behavioral simulation as such. But my point is that AI reaches far beyond simply simulating the actions a human would make. Natural language processing is great for HCI purposes and so on, but genetic algorithms are an AI concept and are used within distinctly non-human contexts more often than not.
This article is a thumbsucker for technical illiterates, and an insult to Slashdot's readership.
Just because an article isn't techie-savvy doesn't mean the content is unworthy of consideration. The project is in fact an interesting and novel one, and the article at least brings the efforts of a relatively obscure project to the public.
My AI faq [robotwisdom.com] gives infinitely more perspective than this [short, introductory, made-for-the-masses] BBC pap, on the important questions.
I looked at your page and found four things:
1) Vast quantities of ego masturbation
2) Same stuff I can find in any AI book worth a damn
3) Same stuff I can find in 20 minutes on Google
4) Even vaster quantities of ego masturbation
Pardon me if I'm unimpressed.
It is. I was [sorta] one of those AI guys.
Actually, you're wrong. I'm sitting here in the Knowledge Discovery Lab at NCSU, and I can tell you that AI is almost ubiquitous in the high-tech industry as a whole. It is used in incredbly complex data visualizations, like those researched by Dr. Healey in the next room from me. It's used in network security, as in several projects by Dr. Doyle down the hall. It's even used in molecular biology, like Dr. Bahler's work. AI isn't just about making things act like humans. That's only good for human-computer interaction (Dr. St. Amant, my advisor, also just down the hall) and entertainment (Dr. Young, two buildings down). AI is about allowing computers to use concepts approximating human reasoning to solve problems, and those problems range from telling stories to determining how proteins fold.
Actually, I was in Mimesis for a semester, and it's none of that, really. The only basis that I'm aware of was that UT can be modded heavily without spending thousands on licensing the engine. They've got a LISP server they connect to that manages all the game logic and the client is actually mostly a bunch of basic mod stuff, like tourists in the aquarium that wander around and form groups (which I coded) and various HUD changes and so on. All the real AI is developed on the remote controller, so it doesn't require direct modification of the UT engine.
Donkey Kong games aren't a dime a dozen? A Bioware-developed SW game is hardly something to ignore. Fable is hardly vaporware, as screens and trailers are everywhere now. You cut down games because you don't know what they are, yet cite a series of games you've never even seen, much less played. Have fun with all of your sequels of sequels of sequels that you've never even played yet but will OBVIOUSLY be so much better and more original than every Xbox game out there.
I'm currently enrolled at NCSU, so I may be able to clear things up a bit... I'm not entirely sure, but I think it was a professor (Dave McAllister, who taught my Graphics course a couple semesters ago) who was doing research with a company called LIPSinc (http://www.lipsinc.com). Their demos show promise, but it's still far from realistic. Hope this helped. PHoRD