Although we have long been talking about computers as artificial brains, actually it has always been a metaphor. Although less and less. IBM has managed to implement a system of 500 artificial neurons with phase change technology they want to simulate the performance of organic brains. In a way that never before had gotten.
For years, scientists have tried to imitate the versatile computing capabilities of neurons. Today, it seems that is the only way to take a step beyond artificial intelligence projects that are underway and that promise to change the world.
Imitate the brain …
They have been taking small steps. IBM recently got itself create memory chips phase change at a similar price of the current RAM. The phase shift is a technology that we’ve been talking a long time that stores data in non – volatile, is not disappear when you turn off the device that reside in the structure and in its relational pattern.
In essence, this also works the brain. The information resides in different firing patterns of neurons and new connections are developed. Therefore, the challenge is to combine complex nanostructures with an acceptable energy consumption, common materials and stochastic behavior (i.e. relatively random). Without these characteristics, it did not seem possible. And that is precisely what IBM has achieved.
… To develop a true artificial intelligence
IBM new neurons try to imitate not only the operation of natural cells, but its structure. Like them, they have input (dendrites), membrane, soma and axon (output). To replace the membrane, use a GST artificial neurons cells, so called because they are made of antimony, germanium and tellurium. This material turns out to be the main active component of rewritable optical disks and is what allows the phase change.
You may also like to read another article on iMindSoft: Technology needs more women for a new golden age
Furthermore, using this technology they have not only succeeded in creating a scalable nanosystem but have been able to imitate the “stochastic behavior” of real neurons. In real brains, each neuron is unique and will develop progressively, so the system always introduces some randomness in the results.
Over the years we have discovered that create genuine artificial intelligence requires joint development of software and hardware. That is why this step is so important because it opens the door to the next generation of artificial intelligence.