Could technology make your brain redundant?
12th Nov 2011 | 12:00
Meet the computer that could outperform the human brain
Could technology make your brain redundant?
Back in June, Fujitsu's K Computer became the fastest supercomputer on Earth - so fast, in fact, that it outperforms the five next fastest supercomputers combined.
It can process an astonishing eight quadrillion calculations a second (eight petaflops), and Fujitsu had hoped that by 2012 it would have cracked the 10-petaflop barrier.
That would have been a significant achievement for several reasons. The first is that it shows just how quickly technology is progressing - as recently as 2006, scientists were getting excited about supercomputers cracking the one-petaflop barrier.
The second is that some pundits believe 10 petaflops is the processing power of the human brain - so on that basis, the K Computer will outperform the human brain within months.
Will our PCs soon be smarter than we are? Is technology telling us that our time is up?
Fujitsu's K Computer won't replace humanity any time soon, and not just because it can't climb stairs.
The K is enormous: at the time of writing its 68,544 CPUs require some 672 computer racks, and more are being added. By the time the project is complete, the K Computer will consist of more than 800 computer racks. That will take performance from 8.2 petaflops to in excess of 10 petaflops - the equivalent of linking one million desktop PCs.
Delivering that performance uses a lot of energy, and while the K Computer has been designed with efficiency as a priority, that's still relative; powering 600,000 CPUs uses enough energy to power 10,000 houses, and Fujitsu's annual electricity bill for the machine is likely to exceed $10,000.
By comparison, the human brain operates on 20-40 watts, and it's much more powerful than the K Computer. It seems the estimate of 10-petaflop processing power was on the low side: the consensus is that our brains are 10 times more powerful than that, delivering at least 100 petaflops, while some estimates are as high as 1,000 petaflops.
Even if the lower estimate is the right one, computers are some way behind: in June 2011, the combined processing power of the world's fastest 500 supercomputers was 58.9 petaflops - or half a brain.
That's not the only area in which computers are lagging behind. Where the human brain fits, handily, into a human skull, supercomputers need considerably more room. The K Computer's planned 800 server racks may be on the large side, but supercomputer power isn't something you can achieve in something significantly smaller.
In 2009, the US defense technology agency DARPA issued a challenge to the IT community, asking it to build a petaflop supercomputer small enough to fit into a single 19-inch cabinet and efficient enough to use no more than 57 kilowatts of power - tiny amounts by computer standards, but still enormous compared to our brains. Nobody has built it yet.
While IBM predicts that a human-rivalling, real-time supercomputer may well exist by 2019, it also predicts that the computer would need a dedicated nuclear power station to run it.
Thanks for the memory
We know roughly how much processing power our electronic brain would need to have, but what about storage? Our brains don't just process data; they store enormous amounts of it for instant access. How many hard disks would we need to emulate that?
The answer depends on where data is actually stored in our brains. If it's stored in our neurons, with one neuron storing one bit of information, then our brains should be able to store 50 to 100 billion bits of data, which works out at around five to 10 gigabytes.
However, if the brain stores data not in the neurons but in the tens of thousands of synapses around each neuron, then our brain capacity would be in the hundreds of terabytes, possibly even petabytes.
That assumes, of course, that the brain stores data like computers do, which it probably doesn't, and that we can measure memories in the same way we measure digital data, which we almost certainly can't. For example, when we store a memory, do we store the whole thing, or is it truncated? Does our brain run data compression routines, and if it does, how lossy are they?
Some neuroscientists are studying that very thing. In February, neuroscientist Ed Connor published a paper in the journal Current Biology with the catchy title A Sparse Object Coding Scheme in Area V4.
In it, Connor described how the brain compresses data in much the same way JPEG compression reduces the size of photographs: while our eyes deliver megapixel images, our brain prefers smaller sizes and concentrates on storing only the key bits of information it needs to recall things correctly in the future. Connor dubbed the results '.brain files', and said in a statement that "for now, at least, the '.brain' format seems to be the best compression algorithm around."
The more we learn about the brain and its systems, the more complicated it appears. In December, Duke University researcher Tobias Egner found evidence that the brain's visual circuits edit what we see before we actually see it, in a model called predictive coding.
Predictive coding reverses our view - no pun intended - of how our visual circuits work. Whereas previously it was believed that our brains processed the entire image with increasing levels of detail, sending data up a 'neuron ladder' until the entire image had been recognised, predictive coding suggests that our brains first take a guess at what they expect to see, then the neurons work out what's different from that guess.
The predictive coding research demonstrates one of the problems with mimicking the human brain: we're not entirely sure what it is we're mimicking, and the more we find out the harder the task becomes.
Thore Graepel is a principal researcher with Microsoft Research in Cambridge, where he specialises in machine learning and probabilistic modelling. "I don't think people are even entirely sure what the computational power of the brain actually is," he told us.
"If we take the standard numbers like 100 billion neurons in the brain, with 1,000 trillion connections between them, then of course if you just transfer that to bits at a certain resolution then you get some numbers - but the differences between the architectures of the systems are so great that it's really hard to compare. In some senses we're competing with five million years of development."
Competing with millions of years of evolution is difficult and expensive. So why not cheat?
Imitation of life
When the Los Alamos National Lab unveiled its petaflop-scale supercomputer Roadrunner in 2008, there was much speculation about its ability to drive in rush-hour traffic - or at least its theoretical ability, as the 227-tonne machine was a bit big to stick in a hatchback. It turns out that just mimicking what we do in traffic requires petaflop-scale processing, which simply isn't portable enough given today's technology.
That doesn't mean computer-controlled cars don't exist yet, though. They do, and they've been bumping around a Californian air base. The DARPA Urban Challenge pitted 11 computer-controlled cars against one another. There were 89 entrants, but only 35 were accepted; of that 35, only 11 made it through a week of pre-race testing.
The 11 competitors were challenged to navigate a range of urban environments that included moving traffic. Six vehicles successfully navigated the course, and the winner was a Chevrolet Tahoe SUV from Carnegie Mellon University in association with General Motors.
The SUV, dubbed Boss, didn't have a petaflop-scale supercomputer, but it did have GPS, long and short range radar, and LIDAR optical range sensors. Those extra senses enabled it to achieve human-like performance without requiring a human-like brain.
There's another, even cheaper way to get human-style performance from a computer: use humans. That's exactly what Facebook did to build its enormous facial recognition database.
By its own account, Facebook's users add more than 100 million tags to Facebook photos every day. By relying on those tags and looking only at possible matches from a user's social circle, Facebook's automatic facial recognition system doesn't have much heavy lifting to do.
That said, facial recognition software is improving rapidly: in August, Carnegie Mellon University researcher Alessandro Acquisti demonstrated an iPhone app that can take your photo, analyse it with facial recognition software and display your name and other information.
In a few years, Acquisti predicts, "facial visual searches will become as common as today's text-based searches", with software scanning public photos like Facebook profile pictures and using them to identify individuals in crowds.
Accuracy is currently around 30 per cent, but the bigger your database and the more processing power you can throw at the problem, the better the results you'll get. It isn't hard to imagine a Kinect-style sensor that knows who you are, what you're doing and what you're trying to ask your computer to do.
"Having additional sensors is a very powerful technique to get around problems," Graepel says. "For example, in our recent development with Kinect, researchers have spent decades looking at stereo vision: the idea of using two cameras to look at a scene and then using the disparity to infer depth and do the processing of the 3D scene. That was very brittle and required a lot of processing power - and then the Kinect sensor came along and you can do much more.
"But the basic advance was on the sensor side, not on the computational side. Similar examples would include RFID tags: if an object is tagged, you don't need to point a camera to see what it is; you can read the tag and find out what it is. Sensors can make computers appear to be more intelligent, or to emulate intelligence in a way that can be extremely useful."
AI and brain-machine interfaces
Sensors working overtime
Sensors can give computers all kinds of useful input. An electronic tongue developed by the Group of Sensors and BIOSensors at the Universitat Autònoma de Barcelona could help wine makers discover defects in their products, while the Israel Institute of Technology's electronic nose can sniff a person's breath and detect the chemical signature of a cancerous tumour.
Kinect sensors add gesture recognition, depth perception and voice recognition technology. Kinect's voice recognition, like the voice recognition in Windows and various speech applications, works reasonably well, but we're still a long way from the Hollywood vision of super-intelligent computers that can carry out conversations with people.
As Microsoft's Natural Language Processing group points out on its blog, "It's ironic that natural language, the symbol system that is easiest for humans to learn and use, is hardest for a computer to master." Enter Watson, the IBM supercomputer that managed to defeat some human contestants in the game show Jeopardy.
IBM sees Watson as a computing breakthrough - a system capable of understanding natural language and responding in kind. Watson's game show career was just a publicity stunt; the technology is now being trialled in the healthcare industry.
As IBM puts it, "A doctor considering a patient's diagnosis could use Watson's analytics technology, in conjunction with Nuance's voice and clinical language understanding solutions, to rapidly consider all the related texts, reference materials, prior cases, and latest knowledge in journals and medical literature to gain evidence from many more potential sources than previously possible. This could help medical professionals confidently determine the most likely diagnosis and treatment options."
The 80-teraflop, 2,880-core Watson isn't the only visible result of the rapid advances in natural language processing. Microsoft and Google both offer online translators that do a decent job of translating web pages in various languages, and those translators also power impressive smartphone translation apps.
Microsoft used the technology to translate nearly 140,000 Knowledge Base articles into Spanish in 2003, and has since added a further eight languages including Japanese, French and German.
Betting on brains
Understanding and processing language is impressive, but being able to do it to human standards is tough - and we have a test to measure a computer's conversational skills.
The Turing Test, based on a 1950 paper by Alan Turing, is based on the belief that one day computers will be smart enough to make us think they're human. All a computer needs to do to pass the test is to fool the judges into thinking it's human, and so far no computer has managed it.
The test is the subject of a famous $20,000 bet between futurist Ray Kurzweil and Lotus founder Mitchell Kapor. Kurzweil believes that a computer will pass the test by 2029, while Kapor thinks his money's safe.
Kapor argues that the brain isn't a computer. "The brain's actual architecture and the intimacy of its interaction, for instance, with the endocrine system, which controls the flow of hormones, and so regulates emotion (which in turn has an extremely important role in regulating cognition) is still virtually unknown," he writes.
"In other words, we really don't know whether in the end, it's all about the bits and just the bits. My prediction is that contemporary metaphors of brain-as-computer and mental activity-as-information processing will in time [be] superseded".
As Kapor points out, carrying out a conversation is far from simple. "While it is possible to imagine a machine obtaining a perfect score on the SAT or winning Jeopardy, since these rely on retained facts and the ability to recall them, it seems far less possible that a machine could weave things together in new ways or to have true imagination in a way that matches everything people can do," he says.
One of the most interesting areas of research is in brain-machine interfaces. Some interfaces can control prosthetic limbs - DARPA has spent five years and more than $100million developing an extraordinarily clever and life-like robotic arm that will be controlled by a microchip inserted in the brain.
The University of Maryland's Brain Cap uses an EEG cap to control similar devices without the need to implant anything, while others prove that sometimes, two heads can be better than one.
Researchers at Columbia University have created a device called C3Vision - short for Cortically Coupled Computer Vision - that uses an electroencephalogram cap on a human user's head to track brain activity. The user is then shown 10 images per second and asked to look for abstract things that computers have a tough time processing, such as things that look 'strange' or 'silly'.
Inevitably there's a military angle to the technology: C3Vision has already been used to scan satellite images to look for surface-to-air missiles, achieving results that no human or computer could manage alone.
DARPA is also working on other kinds of brain-machine interfaces: its extraordinary robotic arm, developed over five years at a cost of more than $100million, will be controlled by a microchip embedded in users' brains. Programme manager Geoffrey Ling says the arm is "truly transformative - just like the arms each of us has."
The arm has been developed to help soldiers who have been injured in Afghanistan, but could ultimately benefit stroke victims, quadriplegics or anyone else who has lost the use of an arm. Clinical trials start later this year, and the arm could be in use within five years.
IBM's Deep Blue famously beat Garry Kasparov at chess, but was its success due to intelligence or just sheer processing power? As Computer History Museum historian Dag Spicer recalls, Deep Blue team member Feng-hsiung Hsu didn't believe that the chess-playing computer was intelligent.
"Assume that we were playing Kasparov at the World Trade Center and 9/11 hit," he told Spicer. "Kasparov would have run like hell. We would have run like hell. Deep Blue would just sit there, computing."
"Chess has been solved by a very brute force approach, by enumerating all the possible moves and countermoves that can be made and building up a game tree to find the optimum move," Graepel says. Deep Blue used a combination of sheer processing power and training from chess grand masters, but the techniques that worked so well in chess don't work so well in the Chinese game Go.
Creating a computerised Go master is one of the goals in artificial intelligence research, and progress was slow: human players would take on enormous handicaps and still thrash the very best computerised opponents.
"In computer Go, the situation was stagnant for many years and people were trying to use the same techniques that had worked for chess," Graepel says. "Then a new technique was invented under the name of Monte Carlo tree search."
By using random samples of games to inform the computer's decisions, Monte Carlo delivered better results than chess-playing algorithms - and it seems to scale. "The problem with previous techniques was that if someone gave you a computer that was 10 times faster, the software wouldn't run better," Graepel says. "But with Monte Carlo tree search, a faster computer made the software considerably stronger."
Thanks to this, "the Go community has managed to create programs that are competitive with good amateur Go players - not professional ones, but good amateurs."
If there's a common thread across the examples we've seen so far, it's that mimicking or beating humans in just one field is difficult enough. Creating a machine that rivals us in multiple areas is even harder.
Comparisons between supercomputers and human brains are somewhat misleading: with the exception of dedicated brain simulators like IBM's C2 and the Blue Brain Project, supercomputers aren't trying to mimic what's in our heads: they're using their power to work on things minds can't, like climate modelling, nuclear warhead simulation and aerospace engineering.
Introducing the superbrain
When it comes to intelligence, results are all that matters - so if a computer-controlled car can complete a circuit, it makes no difference whether it's the result of enormous computing power or good GPS software and sensors. By specialising, we can get what we want much more quickly than by trying to mimic millions of years of evolution - most of which is unique to humans and unnecessary or undesirable in a computer.
As Hugo Award-winning novelist Charlie Stross writes on www.antipope.org, "We may want machines that can recognise and respond to our motivations and needs, but we're likely to leave out the annoying bits, like needing to sleep for 30 per cent of the time, being lazy or emotionally unstable, and having motivations of its own. I don't want my self-driving car to argue with me about where we want to go today. I don't want my robot housekeeper to spend all its time in front of the TV watching contact sports or music videos."
The idea of intelligent artificial humans makes for fun sci-fi, but in the real world the concept seems more like Maslow's Hammer: the idea that, when all you have is a hammer, everything looks like a nail. Perhaps we've already built an intelligent machine, but because it doesn't look like a Terminator we haven't noticed.
"We already have a huge, distributed intelligence out there in the form of the internet, and maybe it's that our individual brains can't see the scope of what we've already built here," Graepel says. "There are two layers to it. There are all those connected computers, which have lots of processing power, and when they all get connected they have even more; and you can also view the internet as a mechanism to create a global human intelligence, where we have these people collaborating and creating what you might call a superbrain."
Factor in the 'internet of things' - devices and sensors of all kinds connected to the internet - and things start getting really interesting. "You need three things," Graepel says. "Sensors, actuators and the processing power that connects them. These three things can become ubiquitous."
We're already seeing some of that in the form of cloud computing, like the smartphone apps that take voice input, send it to servers for processing and return the results instantly. Imagine the same technology with more sensors, more miniaturisation and access to more kinds of data and you've got what Intel calls a 'digital personal handler" - a virtual assistant in a device wired into your glasses that "would see what you see, constantly pull data from the cloud and whisper information to you - telling you where people are, where to buy an item you just saw, or how to adjust your plans when something comes up."
Intel calls it "collaborative perception", with sensors embedded in everything around us. "For example, an intelligent home management system should not only be able to recognise who is at home to determine what entertainment content to display, but also their mood so as to determine appropriate tone of voice, volume and lighting levels," Intel says.
Such empathy requires multiple technologies: sensors, prior learning, cloud data and even the odd bit of human input. What Intel's describing isn't an intelligent machine; it's an intelligent network. In isolation a device can't hope to rival the human brain, but if you connect it to the cloud it becomes a node in an incredibly powerful network, a network with multiple eyes and ears, multiple processor cores and effectively unlimited storage.
On their own, our devices are just devices. Connected, they form a brain the size of the planet.
Liked this? Then check out In pictures: Fujitsu K - world's fastest supercomputer
Sign up for TechRadar's free Week in Tech newsletter
Get the best tech stories of the week, plus the most popular news and reviews delivered straight to your inbox. Sign up at http://www.techradar.com/register