IBM announced the microchip based on the human brain

How to subtly recreate human fragile brains by synthesizing solid metals and silicon?

IBM said that they have found a solution and demonstrated it by building - verifying two 'cognitive ' microchip chips created based on the inspiration of the human brain.

In the mammalian brain, neurons exchange chemical signals with each other through small gaps called synapses. A long 'tail' of neurons - axons, sends signals from many of its terminals to receptors of other neurons - tree-shaped branches to collect them.

Each silicon chip simulates IBM's brain a few millimeters in size and organizes into a network with 256 parallel wires - representing the branch of the "nerve cells" through the right corner by another rope in the axon.

"Neural joints" are 45-nm transistors that connect interlocking wires and act as the memory of the chip, a chip with 262,144 internal synapses and 65,536 other components. Each electrical signal through a synapse consumes only 45 picajoules / 1000 of conventional computer chips.

Because neurons and synapses are closely interconnected, the parts of the hardware responsible for computing and memorization of IBM chips are also more closely linked to those of conventional computer chips. often.

With regular computers, memory is on the edge of the processor, but in new chips - synapses - memory and microprocessors - nerve cells are stacked on top of each other, so they are Does not consume much energy when circulating electric current.

The 'cognitive computing' microchip chip was created based on the inspiration of the human brain. ( Source: Newscientist.com)

This means that these chips can handle multiple processes more efficiently than conventional computers.

In preliminary tests, these chips were able to play Pong's game, control an authentic car on a racecourse and identify an image or digit on the screen.

These are all common computer tasks that can be completed, but these new chips are able to manage and complete them without a specialized program for each job. These chips can also "learn" how to complete each task if 'trained'.

In particular, by connecting multiple chips together, Dharmendra Modha of IBM Research - Almaden in San Jose, California, hopes to build a supercomputer the size of a shoe box, with 10 billion neurons and 100 trillion synapses consume only a few hundred kilowatts of electricity.

This should be noted - when a standard computer uses a few hundred watts, a supercomputer like IBM's Watson uses only a few hundred kilowatts. In contrast, extremely efficient human brains are estimated to have 100 billion neurons and have at least 100 trillion synapses but consume more than 20 watts.

Kwabena Boahen of Stanford University, California said: scale is one of the problems in the mind of this study. By the time the chips contain as many synapses as the human brain, it will lead to difficulty recognizing this achievement compared to the chips in computers.

The work is funded by the US Department of Defense's Research Projects Support Agency (DARPA) to create computers that can replace the human brain.