News & Analysis
Comment
BobsUrUncle
@defendor: You're the idiot. Come up with a better plan, genius.
KeithSchaub
Panel ponders many-core ICs tripping 'the singularity'
Peter Clarke
3/28/2012 12:09 PM EDT
Computers already smarter than humans -- at specific things
Imagination's Oliver added that there are many examples in human brains of linkages to the body that help drive the behavior. "Can you have intelligence without a body?" he asked adding that if we wish to see the advent of the singularity perhaps we should look for it in robots.
The audience engaged with the panel arguing on the one hand that megaflops were not what is needed to approach human intelligence and that hardware is the easy part on the other; that the missing element is software.
Another member of the audience asked what is the application for such levels of performance, apart from creating an automaton. Jem Davies, ARM vice president of technology, came back with the response that in specific domains you want computers to do things that humans cannot. Laser eye surgery is now done by a machine, he said, because it is more precise and capable than any human.
This led the panel on to a discussion of the Turing test and whether supercomputers had yet been able to meet it. The test, proposed by Alan Turing, is that if a human judge, when devoid of visual and other cues, cannot tell the difference between talking to another human being and talking to a machine, then the machine is effectively intelligent.
Imagination's Oliver argued that the definition of Artificial Intelligence seemed to change so that it encompassed those things that computers are not yet capable of, something more akin to an Arthur C Clarke definition of magic. As soon as computers do become capable of a function, for example speech recognition, that task gets reclassified as not being part of intelligence.
From the floor it was asked if the known inefficiencies of multicore arrays for many tasks, were a limitation that would prevent the advent of the singularity. Oliver said there is no doubt that parallel processing is the best way to simulate or recreate brain-like thinking. Computers just happen to be good at only a few tasks such as high-speed numerical processing, that humans are not so good at.
ARM's Davies admitted that general purpose GPU type processing tends to favor particular classes of problem such as computational image processing "but we should not be limited by our imagination," he said. He took a build-it-and-they-will-come position. "I don't need to know what the killer application is going to be. Human ingenuity will find a way to use the technology." Intel's Dubey also argued in favor of the hardware approaches we have today. "It is not a system problem. It's a programming model problem." Dubey qualified that by saying in an email that the problem hinges on a better algorithmic understanding of the human brain.
Imagination's Oliver added that there are many examples in human brains of linkages to the body that help drive the behavior. "Can you have intelligence without a body?" he asked adding that if we wish to see the advent of the singularity perhaps we should look for it in robots.
The audience engaged with the panel arguing on the one hand that megaflops were not what is needed to approach human intelligence and that hardware is the easy part on the other; that the missing element is software.
Another member of the audience asked what is the application for such levels of performance, apart from creating an automaton. Jem Davies, ARM vice president of technology, came back with the response that in specific domains you want computers to do things that humans cannot. Laser eye surgery is now done by a machine, he said, because it is more precise and capable than any human.
This led the panel on to a discussion of the Turing test and whether supercomputers had yet been able to meet it. The test, proposed by Alan Turing, is that if a human judge, when devoid of visual and other cues, cannot tell the difference between talking to another human being and talking to a machine, then the machine is effectively intelligent.
Imagination's Oliver argued that the definition of Artificial Intelligence seemed to change so that it encompassed those things that computers are not yet capable of, something more akin to an Arthur C Clarke definition of magic. As soon as computers do become capable of a function, for example speech recognition, that task gets reclassified as not being part of intelligence.
From the floor it was asked if the known inefficiencies of multicore arrays for many tasks, were a limitation that would prevent the advent of the singularity. Oliver said there is no doubt that parallel processing is the best way to simulate or recreate brain-like thinking. Computers just happen to be good at only a few tasks such as high-speed numerical processing, that humans are not so good at.
ARM's Davies admitted that general purpose GPU type processing tends to favor particular classes of problem such as computational image processing "but we should not be limited by our imagination," he said. He took a build-it-and-they-will-come position. "I don't need to know what the killer application is going to be. Human ingenuity will find a way to use the technology." Intel's Dubey also argued in favor of the hardware approaches we have today. "It is not a system problem. It's a programming model problem." Dubey qualified that by saying in an email that the problem hinges on a better algorithmic understanding of the human brain.
|
|
|
|
|
Navigate to related information
defendor
3/29/2012 8:42 AM EDT
Wondering how they haven't all really solved the problem with these great Geniuses thinking about it.
Title should read: "self-procalmed multi-core experts panel don't know know jack about artifical intellience".
switching to multi-core design is really an admission of the intellectual bankrupcy... well.. we can't really figure out how to make a better microprocessor, its too acedemically difficult to think about with the types of people we hire these days, so let's just do the obvious thing and plunk down as many of them on a chip as we possibily can fit.
Sign in to Reply
PJames
3/29/2012 6:13 PM EDT
Do you have something else in mind?
The amount of parallelism within a single thread is often fairly limited and modern processors with multiple issue and speculative execution are bumping up against diminishing returns. If you go to multiple threads, it simply becomes a tradeoff of whether it is more efficient to make a single "core" execute more and more threads or simply replicate the core.
More transistors means more parallelism. If the basic model of computing as sequences of logical and arithmetic functions is retained, one can carve up that parallelism at different levels but the results remain largely just an issue of optimization rather than some radically better vision.
Sign in to Reply
defendor
3/29/2012 8:45 AM EDT
This guy especially sounds like an idiot: Pradeep Dubey.
Sign in to Reply
BrainiacV
3/29/2012 10:33 AM EDT
I've always argued along the lines of "We can build planes but we can't build birds." We achieve the same function, but through different means. Machine intelligence will be different than human intelligence. Planes don't fly with the grace of birds. That gray squishy thing in our skulls is not just driven by the interconnections, but is also influenced by the chemicals flowing through it. We forget things, the computers wouldn't. But is that forgetfulness part of how we function? I look forward to true AI, but I don't expect it to be something that can really pass a Turing test, anymore than I expect a plane to perch in a tree.
Sign in to Reply
DrQuine
3/29/2012 10:27 PM EDT
I see a key distinction between "expected / known" and "unexpected / unknown" problem solving. Computers are complete champions in arithmetic - most of us happily hand over such tasks because the computer doesn't get tired, distracted, or careless. On the other hand, open ended inference problems are much more efficiently solved by humans - perhaps using the computer as an information retrieval engine to gather appropriate data. As computers "learn new tricks", that boundary will continue to shift.
Sign in to Reply
Nicholas.Lee
4/1/2012 7:09 PM EDT
Step 1: You make a beefburger, it doesn't moo.
Step 2: Yet make a series of bigger and bigger beefburgers, they still don't moo.
Step 3: You put multiple beef burgers in the same bun, it still won't moo.
Conclusion: It's not the speed, size or amount of beef cells you have, it's how is it connected together that determines whether it will moo or not.
I.e. True AI won't be achieved by having greater processing power alone. We need to solve the hard problem of working out the right architecture first.
Sign in to Reply
moloned
4/2/2012 12:21 PM EDT
The big problem with this argument is that we don't know how the human brain works yet so how on earth can we simulate it? The second problem is that even if we did know how it worked the human brain dissipates about 25W, or about 1 millionth of what an exaFLOP computer would require. This limitation means that very few of these artificial "brains" will be built until our technology can rival the efficiency of the brain. Even assuming Moore's law applied (doubling performance every 18 months) it would take 36 years from the moment we have a "brain-computer" to the moment it was as efficient as a human brain. Moral of the story ... don't hold your breath!
Sign in to Reply
KeithSchaub
4/3/2012 3:49 PM EDT
BTW
We already know how to make birds
http://blog.ted.com/2011/07/22/wow-smartbird-in-the-wild-swarmed-by-seagulls/
Sign in to Reply
BobsUrUncle
4/3/2012 6:00 PM EDT
@defendor: You're the idiot. Come up with a better plan, genius.
Sign in to Reply