The most sophisticated computer today, is the human's brain, and it is the reason why many companies have been trying to mimic it and have failed miserably. However, it appears IBM might become the first to crack the code, as the company is currently working on a chip that is capable of processing information on the fly, similar to the human brain.
The chip is quite impressive, as the power it draws is no more than a hearing aid. If done well, it could excel at problem-solving and processing data at rates that far surpass a supercomputer. The processor is called TrueNorth, and it is designed to mimic the way the brain identifies patterns. It also relies on web of transistors that are densely interconnected, which is also similar to the brain.
The chip is built around what IBM calls electric neurons that are capable of signaling others when a certain type of data passes through the threshold. The neurons work in parallel to organize data into patterns, which suggests the data is changing.
When it comes to the hardware specifications, we understand that TrueNorth comes with a whopping 4,096 processor cores. Furthermore, it is capable of mimicking 256 million synapses and 1 million human neurons. Bear in mind that these are two of the most important fundamentals that are the building blocks of the human brain.
Neuroscientists believe the human brain stores data in patterns of pulses, so IBM set out to mimic this quality of the brain by making the chip encode data as patterns of pulses.
We have to admit that this is a neat experiment, and it is probably the first step in moving away from the traditional chip design we've come to accept.
Already, IBM has tested TrueNorth's ability to work with common artificial intelligence tasks. These tasks include, recognizing images and videos. The chip is so advanced; it should be able to tell when a woman in a video picks up her cellphone or a pair of glasses.
Still, despite the praises TrueNorth is getting in the tech arena, there are some who are still skeptical of its capabilities.
Director of artificial intelligence research at Facebook, Yann Lecun, said he is quite skeptical at IBMs design approach with TrueNorth, and believes it may never outperform the fastest processors available today.
"The chip appears to be very limited in many ways, and the performance is not what it seems," said Lecun in an email dispersed to journalists. He also believes TrueNorth could be inadequate in detecting moving cars and pedestrians.
"This particular task won't impress anyone in computer vision or machine learning," he wrote.
In the coming years, however, we should get an idea of what TrueNorth is capable of doing from what it can't.
