How To Clean Refurbished Bathtubs

The computer that thinks like a cat.

IBM announced the world's most powerful artificial brain. It has 147,456 processors, more than 150 terabytes of memory, requires six thousand tons of cooling, occupies about 4000 square feet and consume megawatts of power. All this produces the equivalent of the brain of a cat very, very slow. More than a hundred times slower than your beloved cat.

gattofili If you are, you might be tempted to say that this shows how intelligent is your beloved flock the retractable claws, which, moreover, have always known this, but go there with caution. The supercars of IBM is a mortgage of the brain, not a copy, and simulations require more and much more computing power than the copies.

It 's a remarkable simulation. At Almaden Research Center in San Jose, California, was in fact virtually replicated the coordinated operation of 1.6 billion interconnected neurons from 9 trillion synapses (complexity compared to a brain of a cat), shattering the record Previous 55 million neurons, which is the brain of a rat. A record obtained by the same team only two years ago. A

when a simulation of the human brain? Probably not between more than two decades. The computing power required is a thousand times higher than the mortgage of a cat, but the pace of progress present such a supercomputer should be available by 2019. Just in time to hunt for Sarah Connor.

The purpose of this research is not to create a breed of cats or the Terminator robot, but under controlled conditions to study the behavior and dynamics of the brain slices without anyone. The real challenge, in fact, is how does the brain to self-organize, it is not enough that there are billions of neurons must also be coherent structure. It is one thing to have a house, another is to have a shapeless pile of bricks in it. The spontaneous organization, created by experience and interaction with the world, is completely different from the current computing model, where everything is preset, and is the reason why organic farming requires a supercomputer to simulate the intelligence dragged by the neck.

But the results are starting to see. A more modest simulation, called Darwin and based on 50,000 virtual neurons, equal to one quarter of the brain of a gnat, was installed in a robot on wheels. As the robot interacted with the surrounding world, its virtual neurons altered their own interconnections, producing spontaneous organization.

The other significant result, which will be presented shortly in detail by the researcher Dharmendra S. Almaden Modha, is the use of supercomputers to decipher non-invasively (without slices to anyone, in fact) the interconnections of the human brain by analyzing images obtained using MRI. It's like an X-ray to make a big pot of spaghetti and figuring out the route taken by each noodle. This work is essential, Modha explains in his blog , to understand how each of us is to represent and process information.

To what purpose do all this? Why is cool, but mostly because today's computers are stupid and unable to handle the ambiguities of the real world and to analyze information based on context, and in an increasingly complex world should be complementary intelligence natural and artificial, as well as complement our muscle strength with that of the machines. And also because, to be frank, the current record holder of intelligence on this planet is proving all too often not up to the task.