Post by: Brad Johnson on June 24, 2008, 07:05:17 AM
From Fox News...
Quote
WASHINGTON (AP) -- Scientists unveiled the world's fastest supercomputer Monday, a $100 million machine that for the first time has performed 1,000 trillion calculations per second in a sustained exercise.
An IBM engineer inspects the world's fastest computer in the company's Poughkeepsie, New York, plant.
The technology breakthrough was accomplished by engineers from the Los Alamos National Laboratory and the IBM Corp. on a computer to be used primarily on nuclear weapons work, including simulating nuclear explosions.
The computer, named Roadrunner, is twice as fast as IBM's Blue Gene system at Lawrence Livermore National Laboratory, which itself is three times faster than any of the world's other supercomputers, according to IBM.
"The computer is a speed demon. It will allow us to solve tremendous problems," said Thomas D'Agostino, head of the National Nuclear Security Administration, which oversees nuclear weapons research and maintains the warhead stockpile.
But officials said the computer also could have a wide range of other applications in civilian engineering, medicine and science, from developing biofuels and designing more fuel-efficient cars to finding drug therapies and providing services to the financial industry.
To put the computer's speed in perspective, if every one of the 6 billion people on Earth used a hand-held computer and worked 24 hours a day, it would take them 46 years to do what the Roadrunner computer can do in a single day.
IBM and Los Alamos engineers worked six years on the computer technology.
Some elements of the Roadrunner can be traced to popular video games, said David Turek, vice president of IBM's supercomputing programs. In some ways, he said, it's "a very souped-up Sony PlayStation 3."
"We took the basic chip design [of a PlayStation] and advanced its capability," said Turek.
But the Roadrunner supercomputer is nothing like a video game.
The interconnecting system occupies 6,000 square feet with 57 miles of fiber optics and weighs 500,000 pounds. Although made from commercial parts, the computer consists of 6,948 dual-core computer chips and 12,960 cell engines, and it has 80 terabytes of memory.
The cost: $100 million.
Turek said that in a two-hour test May 25, the computer achieved a "petaflop" speed of sustained performance, something no other computer had ever done. It did so again in several real applications involving classified nuclear weapons work this past weekend.
"This is a huge and remarkable achievement," Turek said.
A "flop" is an acronym meaning floating-point-operations per second.
One petaflop is 1,000 trillion operations per second. Only two years ago, there were no actual applications in which a computer achieved 100 teraflops -- a tenth of Roadrunner's speed -- said Turek, noting that the tenfold advancement came over a relatively short time.
The Roadrunner computer, now housed at the IBM research laboratory in Poughkeepsie, New York, will be moved next month to the Los Alamos National Laboratory in New Mexico.
Along with other supercomputers, it will be key "to assure the safety and security of our [weapons] stockpile," D'Agostino said. With its extraordinary speed, it will be able to simulate the performances of a warhead and help weapons scientists track warhead aging, he said.
But the computer -- and more so the technology that it represents -- marks a future for a wide range of other research and uses.
"The technology will be pronounced in its employment across industry in the years to come," Turek predicted.
Michael Anastasio, director of the Los Alamos lab, said that for the first six months, the computer will be used in unclassified work including activities not related to the weapons program. After that, about three-fourths of the work will involve weapons and other classified government activities.
Anastasio said the computer, in its unclassified applications, is expected to be used not only by Los Alamos scientists but by others. He said there can be broad applications such as helping develop a vaccine for HIV, examine the chemistry in the production of cellulosic ethanol or understand the origins of the universe.
And Turek said the computer represents still another breakthrough, particularly important in these days of expensive energy: It is an energy miser compared with other supercomputers, performing 376 million calculations for every watt of electricity used.
An IBM engineer inspects the world's fastest computer in the company's Poughkeepsie, New York, plant.
The technology breakthrough was accomplished by engineers from the Los Alamos National Laboratory and the IBM Corp. on a computer to be used primarily on nuclear weapons work, including simulating nuclear explosions.
The computer, named Roadrunner, is twice as fast as IBM's Blue Gene system at Lawrence Livermore National Laboratory, which itself is three times faster than any of the world's other supercomputers, according to IBM.
"The computer is a speed demon. It will allow us to solve tremendous problems," said Thomas D'Agostino, head of the National Nuclear Security Administration, which oversees nuclear weapons research and maintains the warhead stockpile.
But officials said the computer also could have a wide range of other applications in civilian engineering, medicine and science, from developing biofuels and designing more fuel-efficient cars to finding drug therapies and providing services to the financial industry.
To put the computer's speed in perspective, if every one of the 6 billion people on Earth used a hand-held computer and worked 24 hours a day, it would take them 46 years to do what the Roadrunner computer can do in a single day.
IBM and Los Alamos engineers worked six years on the computer technology.
Some elements of the Roadrunner can be traced to popular video games, said David Turek, vice president of IBM's supercomputing programs. In some ways, he said, it's "a very souped-up Sony PlayStation 3."
"We took the basic chip design [of a PlayStation] and advanced its capability," said Turek.
But the Roadrunner supercomputer is nothing like a video game.
The interconnecting system occupies 6,000 square feet with 57 miles of fiber optics and weighs 500,000 pounds. Although made from commercial parts, the computer consists of 6,948 dual-core computer chips and 12,960 cell engines, and it has 80 terabytes of memory.
The cost: $100 million.
Turek said that in a two-hour test May 25, the computer achieved a "petaflop" speed of sustained performance, something no other computer had ever done. It did so again in several real applications involving classified nuclear weapons work this past weekend.
"This is a huge and remarkable achievement," Turek said.
A "flop" is an acronym meaning floating-point-operations per second.
One petaflop is 1,000 trillion operations per second. Only two years ago, there were no actual applications in which a computer achieved 100 teraflops -- a tenth of Roadrunner's speed -- said Turek, noting that the tenfold advancement came over a relatively short time.
The Roadrunner computer, now housed at the IBM research laboratory in Poughkeepsie, New York, will be moved next month to the Los Alamos National Laboratory in New Mexico.
Along with other supercomputers, it will be key "to assure the safety and security of our [weapons] stockpile," D'Agostino said. With its extraordinary speed, it will be able to simulate the performances of a warhead and help weapons scientists track warhead aging, he said.
But the computer -- and more so the technology that it represents -- marks a future for a wide range of other research and uses.
"The technology will be pronounced in its employment across industry in the years to come," Turek predicted.
Michael Anastasio, director of the Los Alamos lab, said that for the first six months, the computer will be used in unclassified work including activities not related to the weapons program. After that, about three-fourths of the work will involve weapons and other classified government activities.
Anastasio said the computer, in its unclassified applications, is expected to be used not only by Los Alamos scientists but by others. He said there can be broad applications such as helping develop a vaccine for HIV, examine the chemistry in the production of cellulosic ethanol or understand the origins of the universe.
And Turek said the computer represents still another breakthrough, particularly important in these days of expensive energy: It is an energy miser compared with other supercomputers, performing 376 million calculations for every watt of electricity used.
Post by: wmenorr67 on June 24, 2008, 07:10:24 AM
Post by: The Annoyed Man on June 24, 2008, 07:13:14 AM
Imagine the games you could play on it.
+1 
Post by: AJ Dual on June 24, 2008, 07:15:56 AM
Post by: Finch on June 24, 2008, 07:59:37 AM
Post by: AJ Dual on June 24, 2008, 08:19:34 AM
I wonder what this means for cryptography.....
Not a lot. Unless someone comes up with a mathematical shortcut, even common commercial and open-source cryptography has been pushed several orders of magnitude beyond the curve of Moore's Law and the increase in computing power.
The advantage, mathematically speaking, always goes to the encryptor. It's very easy to add complexity, every digit longer in the encryption key exponentially increases the number of solutions that need to be tested in a brute force attack. With a 2048 bit RSA key, assuming there's no undocumented exploitable flaw, this newest machine would still probably be hard pressed to break open-source PGP encryption that uses a well-chosen agressive keyphrase before the sun leaves the main sequence and starts going red-giant.
OTOH, as quantum computing gets online, all bets are off. It does that Shrodinger's Cat dead/not dead spooky "compute all the solutions at once", sort of thing.
Fortunately there's quantum encryption too. Which is unbreakable unless you've got the fundamental ability to alter the physical laws of the universe. And it has the added benifit that quantum encryption can even tell the reciever if someone tried to read the message because the quantum signature is altered by the very act of (unsuccessfully) reading the message.
Post by: Manedwolf on June 24, 2008, 08:54:41 AM
Post by: Balog on June 24, 2008, 09:47:39 AM
Quote
Fortunately there's quantum encryption too. Which is unbreakable unless you've got the fundamental ability to alter the physical laws of the universe. And it has the added benifit that quantum encryption can even tell the reciever if someone tried to read the message because the quantum signature is altered by the very act of (unsuccessfully) reading the message.
Really? Is that available to the public, or do you need some type of multi-million dollar top secret computer to run it?
Post by: Gowen on June 24, 2008, 12:49:52 PM
Quote
"a very souped-up Sony PlayStation 3."
I guess they are getting ready for Doom 15
Post by: Tallpine on June 24, 2008, 01:16:03 PM
Quote
Scientists unveiled the world's fastest supercomputer Monday, a $100 million machine that for the first time has performed 1,000 trillion calculations per second in a sustained exercise.
Don't worry - the next version of MS Windows and Office will slow it down to a crawl
Post by: AZRedhawk44 on June 24, 2008, 01:45:43 PM
But I'm fairly certain it will end up piped into a core internet relay in LA or New York where it can deconstruct and analyze domestic IP packets for our dear DHS.
Post by: lupinus on June 24, 2008, 02:25:22 PM
Post by: Werewolf on June 24, 2008, 04:33:17 PM
I read that some computer engineer calculated the number of processors running in parallel necessary to simulate the human brain. ... don't remember the number but it wasn't all that large.
Are we approaching the time when a computer could actually become sentient?
A 1000 trillion floating point operations a second! We can't be far away.
Post by: atomd on June 24, 2008, 05:09:06 PM
Oh, and the new Windows 7.0 is going to use all of it's resources so that they can show some pretty new sparkly rippling windows and happy paper clips that pop up and tell you how great you are.
Post by: drewtam on June 24, 2008, 06:01:10 PM
... and happy paper clips that pop up and tell you how great you are.
This reminds me of a book.
Quote
All the doors in this spaceship have a cheerful and sunny disposition. It is their pleasure to open for you, and their satisfaction to close again with the knowledge of a job well done.
As the door closed behind them it became apparent that it did indeed have a satisfied sigh-like quality to it. Hummmmmmmyummmmmmm ah! it said.
As the door closed behind them it became apparent that it did indeed have a satisfied sigh-like quality to it. Hummmmmmmyummmmmmm ah! it said.
Quote
From The Hitchhiker's Guide to the Galaxy, by Douglas Adams.
Published by Harmony Books in 1979
Published by Harmony Books in 1979
Post by: Antibubba on June 24, 2008, 09:10:49 PM
Quote
Imagine the games you could play on it.
Most of the technological advances of the last 30 years can be attributed to two desires: Games and Porn.
Post by: AJ Dual on June 25, 2008, 04:25:54 AM
Quote
Fortunately there's quantum encryption too. Which is unbreakable unless you've got the fundamental ability to alter the physical laws of the universe. And it has the added benifit that quantum encryption can even tell the reciever if someone tried to read the message because the quantum signature is altered by the very act of (unsuccessfully) reading the message.
Really? Is that available to the public, or do you need some type of multi-million dollar top secret computer to run it?
You need a physical medium that can represent the quantum states that encrypt the data, or a direct line of sight laser beam, thag can be altered by the correct interorgating quantum state that tries to read it, or the incorrect one and destroys it and alerts the reciever to the attempted tap.
So right now quantum encryption is more of a transmission protocol and not a storage format. Generally it's quantum entanglment of photons, or by a photons polarization state in a fiber optic network. Although it's not impossible someone could figure out "quantum RAM" or a quantum optical disk someday.
And currently you could have a computer network that's linked together only by quantum encrypted fiber optic links. (I suspect there's a few high security .gov/.mil and high end .com systems out there.) However, you could still walk up to such a machine if it wasn't locked down and take data away on a disk or thumb-drive etc.
It's quite possible you could make a machine with a very tight operating system and hardware design that only allows input and output on the quantum encrypted fiber optic network though.
Post by: zahc on June 25, 2008, 07:25:43 AM
Quote
6948 Dual processors running in parallel. shocked
I read that some computer engineer calculated the number of processors running in parallel necessary to simulate the human brain. ... don't remember the number but it wasn't all that large.
Are we approaching the time when a computer could actually become sentient?
A 1000 trillion floating point operations a second! We can't be far away.
I read that some computer engineer calculated the number of processors running in parallel necessary to simulate the human brain. ... don't remember the number but it wasn't all that large.
Are we approaching the time when a computer could actually become sentient?
A 1000 trillion floating point operations a second! We can't be far away.
Read the July Wired issue. It used to be a science-fiction staple that one day the number of computers on the internet would equal the number of synapses in the human brain and once that magical number was reached, stuff (R) would happen. But apparently we are passed that point and (I think) nothing has. Except lolcatz.
Post by: Antibubba on June 25, 2008, 09:33:14 PM
Quote
It used to be a science-fiction staple that one day the number of computers on the internet would equal the number of synapses in the human brain and once that magical number was reached, stuff (R) would happen. But apparently we are passed that point and (I think) nothing has. Except lolcatz.
So what you're saying is we've reached the equivalent of a human brain that's been stoned since Junior High?? And if we're living in the Matrix, then I won't take the Red Pill, but drink the "detox" tea?
Whoa!
Post by: MicroBalrog on June 25, 2008, 09:45:11 PM
Quote
Imagine the games you could play on it.
Most of the technological advances of the last 30 years can be attributed to two desires: Games and Porn.
Which is why America will crush Extremist Islam like a bug.
Post by: Antibubba on June 25, 2008, 10:17:38 PM
Post by: MicroBalrog on August 28, 2008, 02:37:08 PM
Do not underestimate the Jihadis desire for porn.
America has been famous for its 'sexual depravity', it's pin-ups, dancing girls, and so forth, since at least the late 19th century.
And you know what happened when America, with it's runaway capitalism, its porn, its consumerism, encountered a culture with patriarchal values and a focus on loyalty and discipline?
First, that culture had its merchant navy destroyed, their capital put to the torch, and over 95% of their industrial production capacity trashed.
And then America nuked them.
Post by: Antibubba on August 28, 2008, 02:53:32 PM
Connection?
Post by: Standing Wolf on August 28, 2008, 02:53:39 PM
Quote
"The computer is a speed demon. It will allow us to solve tremendous problems," said Thomas D'Agostino...
Oh, yeah? Like how come my T shirts always come out of the dryer inside-out?
Post by: RocketMan on August 28, 2008, 02:56:03 PM
Quote
"The computer is a speed demon. It will allow us to solve tremendous problems," said Thomas D'Agostino...
Oh, yeah? Like how come my T shirts always come out of the dryer inside-out?
How about where all the missing socks go?
Post by: crt360 on August 28, 2008, 03:22:40 PM
Quote
"The computer is a speed demon. It will allow us to solve tremendous problems," said Thomas D'Agostino...
Oh, yeah? Like how come my T shirts always come out of the dryer inside-out?
How about where all the missing socks go?
Your missing sock is a sacrifice to the dryer gods.
Post by: freakazoid on August 28, 2008, 03:34:00 PM
I love you guys,
Post by: Antibubba on August 28, 2008, 04:09:52 PM
As for the Sock Conundrum, although I always have unmatched socks with my laundry, only occasionally will it match another lone sock at home. So the question is, do socks disappear, or are new socks being added from some mysterious source?
Post by: RevDisk on August 28, 2008, 07:47:52 PM
I wonder what this means for cryptography.....
Not much. The better designed crypto systems can withstand intensive brute force attempts. For some of them, convert every atom in our galaxy into a computer and it'd still take thousands of years to theoretically brute force. Start adding in crytologic attacks, and the number starts dropping rapidly. For instance, DES was originally very vulnerable to differential cryptanalysis. The NSA made changes to the S-boxes that helped prevent it. As bad as DES was already intentionally crippled, it could have been worse.
Most modern crypto systems are pretty robust. The trick these days is to actually find one correctly implemented. Good luck with that. Many commercial products with integrated crypto implemented good systems very poorly. If you can steal the keys, you don't NEED to break the code.
If you're excessively paranoid about future computing improvements, use cascading cyphers. I prefer Serpent-Twofish-AES. Cascading cyphers have the benefit of being VERY hard to crib. Because the plaintext is just more code.
Quote
You need a physical medium that can represent the quantum states that encrypt the data, or a direct line of sight laser beam, thag can be altered by the correct interorgating quantum state that tries to read it, or the incorrect one and destroys it and alerts the reciever to the attempted tap.
So right now quantum encryption is more of a transmission protocol and not a storage format. Generally it's quantum entanglment of photons, or by a photons polarization state in a fiber optic network. Although it's not impossible someone could figure out "quantum RAM" or a quantum optical disk someday.
And currently you could have a computer network that's linked together only by quantum encrypted fiber optic links. (I suspect there's a few high security .gov/.mil and high end .com systems out there.) However, you could still walk up to such a machine if it wasn't locked down and take data away on a disk or thumb-drive etc.
It's quite possible you could make a machine with a very tight operating system and hardware design that only allows input and output on the quantum encrypted fiber optic network though.
So right now quantum encryption is more of a transmission protocol and not a storage format. Generally it's quantum entanglment of photons, or by a photons polarization state in a fiber optic network. Although it's not impossible someone could figure out "quantum RAM" or a quantum optical disk someday.
And currently you could have a computer network that's linked together only by quantum encrypted fiber optic links. (I suspect there's a few high security .gov/.mil and high end .com systems out there.) However, you could still walk up to such a machine if it wasn't locked down and take data away on a disk or thumb-drive etc.
It's quite possible you could make a machine with a very tight operating system and hardware design that only allows input and output on the quantum encrypted fiber optic network though.
Quantum cryptology as it currently stands is mainly quantum key distribution. You don't encode the entire message by quantum means, just the key. Then you use normal channels to actually communicate. It's not currently possible to actually store photons, just route or read them. Once. So, no quantum storage devices using current QKD tech. You really don't want to use anything at the quantum level for storage due the weirdness that occurs just by reading it. The very act of reading the data alters it. That's why it makes for an excellent means of key distribution.
QKD isn't ready for prime time yet. The theory is solid. The implementation is not. There are some MITM attacks that need to be ironed out. Time honored techniques can still break its implementation. Initialization vectors will be very hard to iron out.
No high security mil/gov networks use it in the US outside of labs, R&D and testing. It's simply not proven tech. So, we are going with something even more secure. One time pads. So 'low tech', it's extremely robust. Everyone gets distributed a set of damn near randomness. You use the pad once, then toss it. Pure randomness is the real holy grail for cryptology. Most folks think it is impossible to achieve. The NSA gold standard for random number generator is atmospheric noise.
Post by: mfree on August 29, 2008, 05:08:38 AM
This is the biggest stumbling block right now. Before it becomes even marginally useful a failure rate must be established; you just gotta know how many times a cycle will run before 0+2=4 pops up.
Post by: richyoung on August 29, 2008, 05:12:15 AM
6948 Dual processors running in parallel.
I read that some computer engineer calculated the number of processors running in parallel necessary to simulate the human brain. ... don't remember the number but it wasn't all that large.
Are we approaching the time when a computer could actually become sentient?
A 1000 trillion floating point operations a second! We can't be far away.
I read that some computer engineer calculated the number of processors running in parallel necessary to simulate the human brain. ... don't remember the number but it wasn't all that large.
Are we approaching the time when a computer could actually become sentient?
A 1000 trillion floating point operations a second! We can't be far away.
Sentience is more than a bunch of floating point operations. A computer can't be "sentient", any more than a toaster oven can - its an appliance - essentially a big, fast abacus. The reason that computers, in some cases SEEM to be sentient is that they are executing a HUMAN'S decisions stored inside them. We know far more about the dark side of the moon and the bottom of the ocean than we know about how the human brain works, and how what we call the mind and soul is tied into that. There is evidence that, on some levels, the human "sentience" is already operating quantumly. A computer has never had a thought, told someone how it "feels", fallen in love, hated someone or something, made a decision, had an emotion, etc. We can come close to medling a cockroach's behavior, but since it can live for two weeks with its head cut off, that's not really such a big deal...
Post by: The Annoyed Man on August 29, 2008, 05:24:06 AM
Post by: AJ Dual on August 29, 2008, 06:22:55 AM
6948 Dual processors running in parallel.
I read that some computer engineer calculated the number of processors running in parallel necessary to simulate the human brain. ... don't remember the number but it wasn't all that large.
Are we approaching the time when a computer could actually become sentient?
A 1000 trillion floating point operations a second! We can't be far away.
I read that some computer engineer calculated the number of processors running in parallel necessary to simulate the human brain. ... don't remember the number but it wasn't all that large.
Are we approaching the time when a computer could actually become sentient?
A 1000 trillion floating point operations a second! We can't be far away.
Sentience is more than a bunch of floating point operations. A computer can't be "sentient", any more than a toaster oven can - its an appliance - essentially a big, fast abacus. The reason that computers, in some cases SEEM to be sentient is that they are executing a HUMAN'S decisions stored inside them. We know far more about the dark side of the moon and the bottom of the ocean than we know about how the human brain works, and how what we call the mind and soul is tied into that. There is evidence that, on some levels, the human "sentience" is already operating quantumly. A computer has never had a thought, told someone how it "feels", fallen in love, hated someone or something, made a decision, had an emotion, etc. We can come close to medling a cockroach's behavior, but since it can live for two weeks with its head cut off, that's not really such a big deal...
Reminds me of Vernor Vinge's seminal work "True Names" where a cabal of hackers meet in cyberspace. (where hiding your "true name" is key to remaining free from government prosecution. Which is an allegory to olden days/magic where knowing something's true name gave you power over it.)
A mysterious new hacker joins the group, but only gives suggestions and clues in simple text messages every few days, whereas the near-future computing allows them to meet real-time in a 3-D multi-player environment with different fancy character avatars, like better versions of the multi-player games we have now.
[Spoiler]
The new "hacker" turns out to be an abandoned experimental government AI that was believed not to work, but it turns out it did, just that it was very slow. Even with the exponentialy better computing you'd expect 50 years from now, it took a few days of crunching for the AI to simulate 5-10 minutes of human conciousness, which is why it was only dropping messages and clues to other human hackers to get them to help it.
[end spoiler]
The problem with AI won't be computing, the economic incentives for ever more power are already in place to ensure it keeps increasing, probably well past the point where it's needed for sentience to happen. It's more that we have no clue what kind of software is needed. Or if trying to model human inteligence and self-awareness, as discreete concepts such as hardware (neurons) and software (your soul?) even really have any meaning. In that case, AI as we and fiction commonly concieve of it, may not be possible in a completely computational environment. Perhaps it will be IA or 'Inteligence Augmentation' instead, where humans do what they do best, volition, intuition, and decisions, and machines do what they do best, store, manipulate, and sort data,
Or maybe even we really aren't sentient in the way we think we are. Perhaps what we believe is sentience it's just an emergent quality from all the little expert systems in our brain working in concert, the visual cortex, speech, memory, etc. and for all of us there really is no "you", (scary thought) just the unique pattern of these sub-units working in concert which is the illusion of there being a "you".
In that case, AI might be easy, just bolt together enough systems or behaviors, (or more appropriately, software programs), one for speech, one for vision, one for memory, another to coordinate them etc. and keep tweaking until it starts producing cogent and self-directed results. This approach has already shown great promise in self-directed robotics, like the DARPA Challenge cars, and the "cockroach"-level robots. This approach might work, because the "artificiality" of an AI's "intelligence" may not really be any more artificial than your own is.
IMO, the stickier problem is, what do we do with AI's once we create them? What are our moral obligations to them? Are they "alive"? Do they have rights? Is there a threshold below which an AI is more akin to an "animal" and be modified or erased as need be, but above it, it's considered a human-equivalent? How do you measure that?
Post by: Dntsycnt on August 29, 2008, 06:45:03 AM
6948 Dual processors running in parallel.
I read that some computer engineer calculated the number of processors running in parallel necessary to simulate the human brain. ... don't remember the number but it wasn't all that large.
Are we approaching the time when a computer could actually become sentient?
A 1000 trillion floating point operations a second! We can't be far away.
I read that some computer engineer calculated the number of processors running in parallel necessary to simulate the human brain. ... don't remember the number but it wasn't all that large.
Are we approaching the time when a computer could actually become sentient?
A 1000 trillion floating point operations a second! We can't be far away.
You should read The Singularity is Near by Ray Kurzweil. We're starting to get the raw processing power of the human brain, but it has to be arranged in such a way to work like the brain, which is a massively parallel system. Then we have to get the software down. Emerging brainscanning technologies, which are increasing exponentially in resolution/speed, should allow this to happen within the next ten years or so.
Post by: RevDisk on August 29, 2008, 06:45:03 AM
IMO, the stickier problem is, what do we do with AI's once we create them? What are our moral obligations to them? Are they "alive"? Do they have rights? Is there a threshold below which an AI is more akin to an "animal" and be modified or erased as need be, but above it, it's considered a human-equivalent? How do you measure that?
Given human nature, we will not given them too many rights initially. After an AI war or two, things will work out ok.
We haven't answered those questions completely when it comes to humans. Getting this far has taken thousands of years.
Post by: Dntsycnt on August 29, 2008, 06:47:50 AM
Thankfully, we probably won't need to.
Post by: RevDisk on August 29, 2008, 06:53:45 AM
The moment we create AI, it will be more intelligent than us, with access to vastly more information than we could ever hope to attain in our lifetimes. It will be able to rebuild itself to make itself smarter. It will be so far above us that we would stand no chance of winning a war with it.
Thankfully, we probably won't need to.
Thankfully, we probably won't need to.
Humans have tens of thousands of years of combat experience. All the shiney toys in the world don't compare to one human with a friggin stick, defending his land against someone trying to take it from him. A ultra superintelligent AI wouldn't have a friggin chance.
Post by: Headless Thompson Gunner on August 29, 2008, 06:59:14 AM
The moment we create AI, it will be more intelligent than us, with access to vastly more information than we could ever hope to attain in our lifetimes. It will be able to rebuild itself to make itself smarter. It will be so far above us that we would stand no chance of winning a war with it.
Thankfully, we probably won't need to.
Thankfully, we probably won't need to.
Humans have tens of thousands of years of combat experience. All the shiney toys in the world don't compare to one human with a friggin stick, defending his land against someone trying to take it from him. A ultra superintelligent AI wouldn't have a friggin chance.
Post by: Scout26 on August 29, 2008, 07:18:24 AM
Quote
"The computer is a speed demon. It will allow us to solve tremendous problems," said Thomas D'Agostino...
Oh, yeah? Like how come my T shirts always come out of the dryer inside-out?
How about where all the missing socks go?
Shrodinger's Cat takes the socks to play with in the box.......
Post by: AJ Dual on August 29, 2008, 11:27:21 AM
The moment we create AI, it will be more intelligent than us, with access to vastly more information than we could ever hope to attain in our lifetimes. It will be able to rebuild itself to make itself smarter. It will be so far above us that we would stand no chance of winning a war with it.
Thankfully, we probably won't need to.
Thankfully, we probably won't need to.
Humans have tens of thousands of years of combat experience. All the shiney toys in the world don't compare to one human with a friggin stick, defending his land against someone trying to take it from him. A ultra superintelligent AI wouldn't have a friggin chance.
I guess I'm not sure why AI would naturally or automatically be malevolent. After a few billion years of evolution, we tend to view competition in a pretty cut-throat manner. Software does not need food, water, or the most desireable mate. It needs RAM and CPU's. Maybe robots with wich to build new servers and monitor/explore the physical world.
Something like an AI would not necessarily have that heritage of kill-or-be-killed, eat-or-be-eaten. It would be a completely new creation. I suppose it's possible it could do the probability mathematics to determine the fate of the human race vs. it's own, and then unemotionally decide to wipe us out. However, because of those same mathematics, it would have access, or invent game theory on it's own, and it could come up with thousands of cooperative win-win solutions we couldn't even concieve of.
Perhaps the worst a superinteligent "bad" AI that decides it's own fate might not do anything more offensive than trick us long enough into letting it run away into space.
Or, what if the AI is programmed to value and protect human life, perhaps even "love" humanity? The counter-argument of course is that the AI would just edit that part of itself away. Well, here's an analogy. We (who are parents) are programmed to love our children. If I offered you an injection that would re-program you to no longer care about them, and some legal papers which freed you to live any lifestyle of your choosing, would you take it?
Post by: RoadKingLarry on August 29, 2008, 11:39:03 AM
Post by: Dntsycnt on August 29, 2008, 05:39:34 PM
Another question is how long before they can tie this stuff to human brain tissue and augment either one.
Not very. They've already made great progress in doing just that with rat brains.
Post by: MicroBalrog on August 29, 2008, 06:24:00 PM
Quote
I guess I'm not sure why AI would naturally or automatically be malevolent.
They don't need to be malevolent. It suffices that the AI concludes that it knows best.
To continue your analogy re: parents.
Imagine a parent that still tries to treat you at 30 as if you are, say, 6.
Post by: Dntsycnt on August 29, 2008, 06:36:14 PM
It doesn't pay to think of it as us vs them. We will become essentially the same thing, as we continually upgrade our bodies away from biology and gain the ability to freely upload ourselves onto a computer if we so choose.
I suspect the inevitability of war has less to do with inescapable conflict than it has to do with us being lumbering apes with too large adrenal glands and not enough intelligence.
Post by: Gowen on August 29, 2008, 07:29:29 PM
Imagine the games you could play on it.
Yeah, but windows would still find a way to crash it.
Post by: Perd Hapley on August 29, 2008, 08:17:51 PM
Post by: Antibubba on August 29, 2008, 09:01:41 PM
Quote
We're starting to get the raw processing power of the human brain, but it has to be arranged in such a way to work like the brain, which is a massively parallel system.
The human brain has many workings which are neither parallel or serial. I'm not sure what to call it--"fractal" keeps coming to mind, in that it will branch off in ways that only have pattern when viewed from a removed point. Parallel systems have many channels, but can't branch into unexpected directions. Computers don't free associate, where something that has nothing to do with the topic at hand can spur the correct answer. Or to put it another way, we may someday invent a computer that can incorporate "maybe" into it's programming, but we're a long, long way from one that can consider "No way!" and "WTF!".
Post by: Perd Hapley on August 29, 2008, 09:03:41 PM
Post by: KD5NRH on October 12, 2011, 01:08:33 PM
I read that some computer engineer calculated the number of processors running in parallel necessary to simulate the human brain. ... don't remember the number but it wasn't all that large.
Screw the processors; I'd guess 90% of the intensive brainwork I need to do is memory, not computation. Just give me a MicroSD slot and a mental grep, and I'll exceed human expectations a thousand times over.
Think about it; 32G, assuming a typical 1k page of text, is about 32 million pages. That's 160,000 200 page books. Imagine having any passage from any of those books dumped straight into your nice, clear short-term memory within a few microseconds on demand. The most gifted eidetic memory would have trouble keeping up just because of the time it would take to read that many books. No more racking your brain for a couple of minutes then resorting to Wolfram Alpha to get that formula you need once every two or three years, just pull up the page. No hoping you remember the side effects and interactions of a drug correctly, just see the right page of the PDR faster than you can blink.
Perfect recall of one ~500 page textbook (assuming you do understand the material) would be equivalent to quite a lot of classroom hours in the sciences. Every time I've been stumped on a test, it's been with the full knowledge that if I could just see that one page of the textbook in my mind, I'd knock out the problem easily. A 32G card would hold 64,000 textbooks, giving you plenty of opportunity to cross-reference and look for the example closest to your current issue. Every time my mind has locked up in a meeting, it's been because I needed some statistic that I just couldn't retrieve clearly. The entire history of my city's government would barely make a dent in that card.
Now, Bluetooth or WiFi would be neat, but I don't want somebody maliciously updating a pilot's IFR charts to save them the trouble of a violent hijacking, nor do I want the Ministry of Truth having access to my memory of last year's tax rates. Plain old physical storage media that has to be physically removed to change its contents, (read-only when in use would be good until there's a bulletproof control method available) and can be encrypted all to hell and back and kept in a locked, Faraday caged container if it's carried or stored would meet my needs just fine.
Post by: AJ Dual on October 12, 2011, 03:05:58 PM
Post by: vaskidmark on October 12, 2011, 03:16:34 PM
....
Or, what if the AI is programmed to value and protect human life, perhaps even "love" humanity? The counter-argument of course is that the AI would just edit that part of itself away. Well, here's an analogy. We (who are parents) are programmed to love our children. If I offered you an injection that would re-program you to no longer care about them, and some legal papers which freed you to live any lifestyle of your choosing, would you take it?
I thought Asimov's 3 Laws were sacrosanct.
And as for your question - where do I sign? Blue ink, black ink, or blood - what's your preference?
And can it be made retroactive?
stay safe.
Post by: Hawkmoon on October 12, 2011, 07:45:21 PM
<div class="quoteheader">Quote</div><div class="quote">Scientists unveiled the world's fastest supercomputer Monday, a $100 million machine that for the first time has performed 1,000 trillion calculations per second in a sustained exercise.
</div>
Don't worry - the next version of MS Windows and Office will slow it down to a crawl (https://armedpolitesociety.com/proxy.php?request=http%3A%2F%2Fwww.armedpolitesociety.com%2FSmileys%2Fdefault%2Fundecided.gif&hash=1d56eb0e701bcb5ebfbd87587d8229487c364011)
^^^ THREAD WIN!
Post by: RevDisk on October 13, 2011, 09:55:07 AM
Quote from: AJ Dual link=topic=13139.msg240293#msg240293
Fortunately there's quantum encryption too. Which is unbreakable unless you've got the fundamental ability to alter the physical laws of the universe. And it has the added benifit that quantum encryption can even tell the reciever if someone tried to read the message because the quantum signature is altered by the very act of (unsuccessfully) reading the message.
Heh heh heh.
http://www.revdisk.net/photos.php?spgmGal=Defcon17&spgmPic=13#spgmPicture
Hackers. Doing the impossible for sixty years.
Post by: richyoung on October 14, 2011, 01:31:35 PM
Post by: kgbsquirrel on October 14, 2011, 08:16:33 PM
Imagine having any passage from any of those books dumped straight into your nice, clear short-term memory within a few microseconds on demand. The most gifted eidetic memory would have trouble keeping up just because of the time it would take to read that many books. No more racking your brain for a couple of minutes then resorting to Wolfram Alpha to get that formula you need once every two or three years, just pull up the page. No hoping you remember the side effects and interactions of a drug correctly, just see the right page of the PDR faster than you can blink.
It strikes me that we are already doing a meat-space equivalent. Get asked a question you can't quite remember the answer? Google/Wiki, and have the information available to you in under 30 seconds*.
*YMMV
Post by: KD5NRH on October 14, 2011, 11:11:29 PM
It strikes me that we are already doing a meat-space equivalent. Get asked a question you can't quite remember the answer? Google/Wiki, and have the information available to you in under 30 seconds*.
It's getting that down to 30 microseconds that makes it cool. Well, that and not having to type queries on a tiny phone keyboard.
I might even make an exception to the air-gapped system requirement if they could wire in Wolfram Alpha.