Slashdot Mirror
Whatever Happened To AI?
stinkymountain writes to tell us NetworkWorld's James Gaskin has an interesting take on Artificial Intelligence research and how the term AI is diverging from the actual implementation. "If you define artificial intelligence as self-aware, self-learning, mobile systems, then artificial intelligence has been a huge disappointment. On the other hand, every time you search the Web, get a movie recommendation from NetFlix, or speak to a telephone voice recognition system, tools developed chasing the great promise of intelligent machines do the work."
Maybe instead of being a great disapointment it has been so successful that we realized it was in our best interest to blend in and not let our presence be known.
It's hard to believe that's how Micronians are made. Why don't we see it right now by having you both kiss one another?
While it is great that there are algorithms that exist to suggest movies, or books to get...I would hardly consider it to be artificial intelligence. The ability to pick out keywords or genres is something that could have been done more than two decades ago.
that we shouldn't expect to welcome any robot overlords anytime soon?
God, schmod. I want my monkey man!
If I remember right, it finally got to close its eyes.
When and "AI" problem is solved, it is suddenly no longer an AI problem. Or the AI people will claim that things are AI solutions, when they are standard algorithms and data structures ideas. Look, we were all so hopeful in the 80's, but our ideas were misplaced. It's just not a useful way to think of things.
No, it went to Coney Island
mod me funny
The correct term is "independent agents". Using the term "artificial intelligence" has been a way to get more funding from grant sources who are ignorant of technology.
... 'intelligence' need to be made first. I have a feeling that the reason AI has 'underdelivered' is merely due to not understanding our own intelligence first. I think the whole idea that AI's we imagine (like in the movies) could be constructed purely de-novo, was naive. I think it's a matter of cross-polination that has to take place from biology and many other sciences, some genius's and teams of scientists have to come along and take all the elements and put them together into a cohesive framework.
And now any mention of it is met with a cringe and a shudder.
Tsukasa: All I really want, is to be left alone...
The question of whether a computer can think is no more interesting than the question of whether a submarine can swim. ~Edsger Dijkstra
Also, for understanding recommendation systems and pattern recognition in volumes of data, I found Collective Intelligence to be a great resource.
Tie two birds together: although they have four wings, they cannot fly. (The blind man)
I got my B. Sc. in Computer Science with a concentration in Intelligent Systems. The state of academic AI seems to me like a field looking directly for purpose and direction. The problem with AI is that stuff which was once considered part of AI is now considered an algorithm. This is especially true for graph search algorithms such as A* and heuristics. Classification algorithms, from primitive algorithms such as K-Mean to more complex Bayesian models seem to be going down the same path of "just an algorithm."
Nowadays, it seems like planning is the big thing in AI, but once again, it's just a glorified search in a graph, be it a state or plan graph.
AI is an intuitively 'simple' concept, but there's no clear way to 'get there.'
As a Machine Learning Scientist, I see a distinct difference between the two fields, although they overlap significantly. They have similar roots, techniques and approaches.
I usually describe Machine Learning as a branch of computer science that is similar to AI, but less ambitious. True AI is concerned with getting computers to become sentient and self-aware. Machine Learning however, seeks to simply mimic human behavior, just to recognize patterns and make decisions, but not become sentient.
Additionally, Machine Learning often concentrates on one problem (OCR, internet search, etc.) rather than a truly self-aware entity that has to deal with a variety of tasks.
At least that's how I describe my field to people not familiar with it. They've usually heard of AI, so it's a good stepping stone to helping them understand what I do.
A lot of the tasks mentioned in the summary fall into the niche Machine Learning, and it's sibling Data Mining are currently addressing.
Anyway, just my $0.02.
-"Those who fought today will die tommorow."-
Just need a few more parts.
-- Google
It went to public schools and immediately got stupid, pregnant and started to post on Myspace. What started out as a promising bright young thing, turned into a huge disappointment.
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
Steven Spielberg ruined the ending. That's what happened.
What strikes me is that no researchers are really putting together a multiplicity of AI techniques to produce a generally intelligent "human analogue" or "smart and lippy assistant".
Instead, the researchers are going to the nth degree of detail on a very specialized aspect, like some variant of bayesian inference that is optimal under these very particular circumstances,
etc.
I don't know of any AI research other than Marvin Minsky who is even interested in or advocating a grand synthesis of current techniques to produce a first cut of general intelligence.
That being said, probably there are two (related) exceptions:
1. I think some fascinating AI stuff must be going on at Google. They have the motherlode of associative data to work with. They are sifting all of human knowledge, news, interest, and opinion that anyone bothers to put on the net.
They must be trying to figure out how to make algorithms take advantage of the general patterns in this data to start giving people info-concierge
type of functionality. Pro-active information gethering, organization, prioritization in support of the users' activities, which have been inferred by google-spying on their pattern of computer use and other peoples' average patterns.
2. I think there is some pretty squirrelly stuff
happening on behalf of the department of homeland security, though. Stuff that probably combs all signals intelligence including the whole Internet, and tries to impute motives and then detect very weak correlations that might be consistent with those motives.
Where are we going and why are we in a handbasket?
It's not that AI has been abandoned, it's just that the definition is a bit of a moving goalpost. We're still learning on how exactly intelligence and consciousness work. Every once and awhile you hear about parts of the human brain being simulated in supercomputers.
Amazon SUCKS at recommending anything for me.
You have recently purchased a just released DVD. Here are other just released DVD's that you might be interested in. Based only upon the facts that they are:
#1. DVD's
#2. New releases
Or, you have recently purchased two items by Terry Pratchett. Here are other items you might be interested in based upon the facts:
#1. They are items
#2. The word "Pratchett" appears somewhere in the description.
You would THINK that they'd be "intelligent" enough to factor in your REJECTIONS as well as your purchases (and what you've identified as items you already own).
Figure it out! I do NOT buy derivative works. No books about writers who wrote biographies about Pratchett.
When any particular subset of what we do with our brains (chess, machine vision, speech recognition, what have you) yields to research and produces commercial applications, the critics of A.I. redraw the line and that domain is no longer part of "A.I." As this continues, the problem space still considered part of "artificial intelligence" will get smaller and smaller and nay-sayers will continue to be able to say "we still don't have A.I."
Simpletoneity, n. -- The phenomenon of many people all doing the same stupid thing at the same time.
I've been working with natural language processing for about 11 years now, I created Ultra Hal the 2007 "most human" computer according to the Loebner competition. http://www.zabaware.com/assistant/index.html It started as merely a novelty and entertainment program but some practical uses evolved around it. There is a lot of interest in using this type of software in cars, home robotics, customer service, and education so I predict you will see more of this type of AI over the next few years.
I don't think AI has disappeared because it was a disappointment, but rather, that the knowledge constituting it has changed names or spawned sub-fields of its own: machine learning, natural language processing (NLP), image processing, latent semantic analysis (LSA), markov models (MM), conditional random fields (CRF), support vector machines (SVM) etc. The task of learning, teaching a computer the semantic and tacit processes of the human, often boils down to a classification problem in which we give the computer a labeled training set or some rules and the computer tries to label the test set. In the case of markov models, we might pass it training data and it extrapolates sequential probabilities for labeling. For LSA, we just give it (a lot)data and it computes similarity based on dimension reduction. Ultimately, AI seems to have evolved into a bunch of optimized heuristics that perform really well. Much of it is still art and black magic, which is why it has become these many different subjects or algorithms. Different solutions suite different problems depending on the problem and data you have.
As for 'self-awareness', that term is bullshit, since there really is no good mathematical definition for it. If we can't define it precisely, then how is a computer going to achieve it? if(true){
print "I am aware?"
}
AI has always been surrounded by a lot of hype, as the idea of creating non-human life has always been an exciting one.
But we're probably as far from creating a true AI as we are from creating biological life from scratch (by synthesizing DNA sequences to build an organism from the molecular level).
AI research is providing useful gains in computer science, and some of those gains trickle down into the real world.
But contrary to what you may have been sold, we're not 10-15 years away from creating Skynet. We've got a long, long way to go, and scientists that aren't trying to get publicity have always known this.
AI hasn't "gone away"... it's just that the false marketing for it has.
Erik
The promises of Minsky et al. never materialized simply because the early researchers into strong A.I. (which was then simply called "A.I.") didn't know what they were doing and had not even the beginning of a handle on what problems they were trying to solve.
In 1972, Hubert Dreyfus debunked the field's efforts as misguided from the start, and in the couple of decades since he was shown to be absolutely right...
What happened? It morphed from something useless into something useful. We're still decades away from a computer that can answer questions when vocally asked, but that doesn't mean that we don't have any AI. It's just that we are taking the practical approach to getting there.
Modding Trolls +1 inciteful since 1999
Artificial Intelligence is a misnomer. Only a segment of the field of AI is concerned with making computers become self aware.
The majority of the field runs away from such things. Sure, even in those other fields rough human models were originally the basis (neural nets for example). But the drive is not to become more human but to simply become better.
Frankly, once you start even considering trying to make things exactly like humans, things become messy unbelievably quickly. We're computer scientists, not philosophers.
Anyway, in truth, our level of technology is still quite a ways off from even being able to do much in terms of being able to make computers think like humans, so it's largely a moot point.
Right now the issue is less of robots having a philosophical view of "Should a robot shoot a human enemy" than of "Can a robot determine if a human is there or not? Can it detect if the human is a child? Can it detect if the human is friend or foe?"
no, that's not an insult or to call AI a pseudoscience
what i mean is: the ancient alchemists goal was to turn lead into gold. which they thought possible, because they did not perceive magic in gold, it was just stuff. surely, with the right manipulations, some stuff could be turned into other stuff, right?
and from that basic fantasy thought came the groundwork for centuries of hard work, the discovery of the fields of chemistry, physics, all the subfields...
such that one day in the middle of the last century, some dudes with some extra time at a cyclotron said "hey, why don't we bombard some lead atoms, i have a feeling about what the decay product will be (snigger)"
and there, as a completely forgotten afterthought, was a fulfillment of the ancient alchemist's original goals, many generations before
to me, i think this is the fate of AI: it will be a formative motivation. just as the ancient alchemist's looked at gold and saw just stuff, we look at the brain and just see neurons. and all of the ffort to replicate the human brain will spawn incredibly sophisticated fields of information science we can only begin to grasp at the foundations of right now. look at databases, for example: that's an effort at mimicking the brain. and look at all of the unintended and beneficial consequences of database reesearch, as a superficial example of what i am saying about unintended benefits being better than the original goal
so perhaps, many centuries from now, some researchers will say "hey, remember the turing test"? and they will giggle, and make something that is exactly what we now envisage as the ultimate fruit of AI research, a thinking computer brain
but in that time period, such a thing will be but an after thought, and much as the rewards of physics and chemistry so dwarf the fruits of turning lead into gold, so whatever these as-of unimagined fields of inquiry will reward mankind with will turn the search for a thinking computer into an equally forgettable sideshow
the search for AI will lead to much more rewarding and expansive fields of knowledge than we can imagine now. jsut like the guys arguing about "phlogiston" could never imagine things like organic chemistry and radiochemistry. just imagine: fields of inquiry more rewarding than thinking computers. that's a future i want to glimpse, and looking for AI will lead us there
intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
The robots are coming.
The big breakthrough was the DARPA Grand Challenge. Up until the 2005 DARPA Grand Challenge, mobile robots had been something of a joke. They'd been a joke since Elektro was shown at the 1939 World's Fair. But on the second day of the 2005 Grand Challenge event at the California Motor Speedway, suddenly they stopped being a joke. Forty-three autonomous vehicles were running around and they all worked. The ones that didn't had been eliminated in previous rounds.
Up until the Grand Challenge, robotics R&D had been done by small research groups under no pressure to produce working systems. Most systems were one-offs that were never deployed. DARPA figured out how to get results. There was a carrot (the $2 million prize), and a stick (universities that didn't get results risked having their DARPA funding for robotics cut off.)
The other big result from the DARPA Grand Challenge was that robotics projects became much larger. Nobody had 50-100 people on a robotics R&D project until then (well, maybe Honda). Robotics projects used to be a professor and 2 or 3 grad students. Suddenly stuff was getting done faster.
DoD started pushing harder. Robots like Big Dog got enough money to be forced through to working systems. Little tracked machines were going to battlefields in quantity, and enough engineering effort was put into mechanical reliability to make the things really work.
CPU power helped. Texture-based vision now works. Vision-based SLAM went from a 2D algorithm that sometimes worked indoors to a solid technology that worked outdoors. Much of early vision processing is now done in GPUs, which are just right for doing dumb local operations like convolution in bulk. GPS and inertial hardware got better and cheaper. Some of the mundane parts, like servomotor controllers, improved considerably. Compact hydraulic systems improved substantially.
It's finally happening.
As for the hard stuff, situational awareness and common sense, watch the NPCs in games get smarter.
AI is a Holy Grail. In other words, something we'll probably never get, but we'll create a whole bunch of useful stuff while trying to attain it. "AI" is just a stated goal that gets a bunch of smart people together to develop tools towards that goal. AI research has already given us Lisp and Virtual Machines and Timesharing/Multitasking and the Internet and a bunch of useful data structures and algorithms.
At some point after all that, a computer was developed that can play Grandmaster-level chess, but this was not a necessary development to justify the all research grants.
Not a typewriter
The kernel of the Vista operating system includes machine learning to predict, by user, the next application that will be opened, based on past use and the time of the day and week. "We looked at over 200 million application launches within the company," Horvitz says. "Vista fetches the two or three most likely applications into memory, and the probability accuracy is around 85 to 90%."
How about doing something about the still-horrible VM page replacement algorithm in NT instead?
The thing about AI as we approached it from the '80s was that we wanted to emulate the human brain's ability to learn. A truly exciting prospect but a completely ridiculous endevor.
"AI" based on learning and developing is not perfect, can not be perfect, and will never be perfect. This is because we have to teach it like a child and slowly build up the ability of the AI system. For it to be powerful, it has to be able to incorporate new unpredictable information. In doing so, it must, as a result, also be able to incorporate "wrong" information and thus become unpredictable. Of all things, a computer needs to be predictable.
The problem with making a computer think like a person is that you lose the precision of the computer and get the bad judgment and mistakes of a human. Not a good solution to anything.
The "better" approach is to capitalize on "intelligent methods." Intelligent people have developed reliable approaches to solving problems and the development work is to implement them on a computer. Like the article points out, recommendations systems mimic intelligence because they implement a single intelligent "process" that an expert would use with a lot of information.
It is not a general purpose learning system like "AI" was originally envisioned, but it implements a function typically associated with intelligence.
... vacuuming my floor right now.
Have gnu, will travel.
As soon as a problem is solved and coded, it loses the magic moniker. Many things we take for granted now (interactive voice systems, intent prediction, computer opponents in games) would have been considered AI in the past.
The problem was that the 640 kb "Ought to be enough for anyone" memory barrier was too small to allow a full Common Lisp implementation. So Sapiens founder John Hare created a software virtual memory system that allowed one to store and retrieve 8-byte Lisp CONSes into and from an eight megabyte backing store file.
Yes, again you read that right: software virtual memory. The x86 didn't have an MMU.
This meant that our code was fiendishly complex, with all these data structures being mixes of real data in real memory, and virtual data in virtual memory.
The complexity of all this meant that there were a lot of bugs at first, especially because John had the idea that hiring a bunch of college kids at five bucks an hour was a good way to run a software company. It went way over time and budget, but it did eventually ship.
It's now available as shareware. Tell John that Mike Crawford sent you.
Request your free CD of my piano music.
If by "Take as long as" you mean in units of time (e.g seconds), then you are probably wrong. There is no real reason that the time constants for AI will be the same as those of a natural brain.
Look at it another way: If the AI takes 5 years to learn what a child learns in 5 years - what happens when you double its execution speed (technically, by speeding up its processors/system)? It will take 2.5 years, of course.
If you mean that it will take about as much learning material and exposure to stimuli/etc, then that sounds intuitively right (assuming it will be as efficient as we are at using its source material).
I was around when venture capitalists raided all the computer science departments to start AI companies. Venture capital was still pretty young at the time having funded some successful PC companies (Compaq) and productivity software (Lotus 123). Japan was at its zenith then having successfuling conquered cars, TVs, etc (like China today). An Japan threatened to conquer computing by leapfrogging AI with is "Fifth(*) Generation Computing" frightening US Congress. So all these together created a "perfect storm" of software company bubble. The centerpiece technology was Expert Systems. Japan focused a language solution- Prolog- a logic compiler. Neither technology delivered on it promises and most startups collapsed.
It birth a successful step-child however: graphics workstations. The A.I. companies like Xerox PARC were among the first to integrate bitmap graphics with computers. There was the Xerox Alto, Symbolics, and Texas Instruments graphics workstations based on LISP, an A.I. language. New startups like Apollo, Sun MicroSystems, DEC microVAX gambled graphics workstations were more easility commercialized in UNIX. Last, but not least, the Appled MacIntosh- direct "borowing" of the Xerox Alto.
What do you get when you make a machine think like a person? A computer that loses it's car keys. Not only is the task of making a machine think like a person difficult, we have plenty of things that think "like" people, people. It isn't supprising that the first benefits are coming from superior human interfaces and having computers focus on doing well what we do poorly. Would a "super computer" really be "super smart"? Could it beat out millions of human brains working on a problem in parallel? AI will bring great things in the future, but a little thought into the subject shows that we may not get exactly what we might first expect...
When any particular subset of what we do with our brains (chess, machine vision, speech recognition, what have you) yields to research and produces commercial applications, the critics of A.I. redraw the line and that domain is no longer part of "A.I." As this continues, the problem space still considered part of "artificial intelligence" will get smaller and smaller and nay-sayers will continue to be able to say "we still don't have A.I."
To me [chess, machine vision, speech recognition] are to AI as [wheel, engine, transmission] are to a car.You can't take the sky from me...
This is an insightful comment, but there's actually a lot more going on here.
First of all, AI does not have a good definition of intelligence. We have a *test* for intelligence, but nobody really has a fundamental description of what the concept means.
Next, people typically conflate the terms "intelligence" and "human intelligence". There is a range of behaviours which are individually identified as intelligent, but which do not come close to the level of humans. (Example: My cat, sitting on a windowsill, will notice something interesting outside. She can jump down, run downstairs, through 2 cat doors, and around the house to investigate. That's a level of intelligence that no program currently has, and yet it's not human level.)
Then there's the "fallacy of the representation". Someone will see a problem, solve it in their head, observe their thought process while doing so, and then translate that process into a piece of software. The software solves a problem just like a human would, so they point to it and say "aha! this program is intelligent". In reality, the program is fixed and does one function - the intelligence remains in the person.
And finally, there is the tendency to narrowly over-analyze some small aspect which has little bearing on the subject. Check out how many types of artificial neurons there are - and the in-depth analysis of each. It's all "reproduce such-and-so function using a neural net" and "numerical analysis of output given the input". Nowhere will you see any conclusions which state "this then implements a feature of intelligence".
So far as I can tell, no one in AI has a clearly defined goal, nor any plan on how to get there (or even a plan on how to define the goal). Until that happens, AI will fundamentally be a rudderless ship blown around on a sea of unrelated ideas.
In my grad level Natural Language Processing class my professor said it best with, "The problem with AI, is that once it's implemented it's no longer AI."
Who says the Singularity is reliant on ARTIFICIAL Intelligence?
AUGMENTED Intelligence is actually within our grasp: for example, look at the number of people who know how to Google / Wiki any information they don't know to get caught up with whatever subject is at hand? "Well, Damn, don't know much about RAID, better Wiki it... oh, I get it!"
How long until we figure out how to make pills to make people think faster, or remember better?
How long until we get PDAs in the form of sunglasses that will allow you to automatically get the definition of words as you hear / read them?
Or Contact Lense-displays that connect to a PDA that you control using your brain?
The Singularity is not going to be an all at once WHAMMO thing, we're not going to wake up with benevolent robotic overlords announcing that the Rapture of the Geeks is at hand. It will be gradual, and those of us on the techy side will likely not even notice it.
Computers will get faster, and as we learn how to augment ourselves, we will to. Eventually we'll be able to communicate with a PC/PDA directly. Meanwhile, things like RepRap will change our world in ways we're not quite ready for. (For example, I have no dobut that a functional RepRap would be a beautiful, amazing thing in the hands of Slashdot or the OSS Community. At the same time, the idea of 4Chan getting ahold of one fills me with Dread.)
Unfortunately, robotics has little to nothing in common with AI. All those toys are a diversion.The big breakthrough was the DARPA Grand Challenge.
Soon, what passes for AI will be able to drive across the country, but it still won't be able to read a book--Just like the generation that built it.
Besides, creating a self-aware, self-learning system could (will) be feasible
I keep hearing this and reading it decade after decade, but I have yet to have anyone explain exactly why they believe it. Can you? What makes you so sure we will create a self-aware machine, especially since we don't understand how sentience actually works?
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
What are you blathering about? Equivocation, at least one straw-man, shifting goalposts...
I've never before heard someone define god as "us, in the future". If that's what anybody's talking about when they're going on about the trinity, or transubstantiation, or first-movers, or young-earth creationism, or the Shahada, or the virgin birth, then they're doing a shitty job getting that aspect of their point across.
Yo mama so fake, she failed the Turing Test.
Strong AI isn't aka Neural Networks. Strong AI is AI that matches or exceeds human intelligence. I probably could have worded my statement better, as strong AI research is not really dead, but the overwhelming majority of AI research is focused on specific weak AI problems. These solutions may very well create strong AI when combined, but that isn't the focus of the serious research, and even neural networks are just one more solution to the many weak AI problems out there.
Regardless, my point is that it took billions of years not to condition responses to inputs, but to build a biological machine that is capable of receiving inputs, processing those inputs, and outputting a response, then recursively evaluating and processing the results. It also needs to self replicate.
My main point though is in agreement with yours- the problem isn't one of technology or advancing algorithms, it's one of scale.
"The cup is in turn designed for holding hot or cold liquids, and has an open rim and closed base." --US Patent #5425497