IBM's New Chip Works Like Your Brain

Written by Alex Zaharov-Reutt      08/08/2014 | 10:12 | Category: GENERAL

IBM has developed a new chip to mimic the function of the human brain that will one day give us truly smart robots - but hopefully not the Terminator.

IBM might just have to be renamed the International Brain Machines company following its creation of a new chip called "True North" that mimics the way human brains work. 

The news is published in the latest edition of Science Magazine which notes in the article abstract that "computers are [still] nowhere near as versatile as our own brains". 

IBM researchers have applied "our present knowledge of the structure and function of the brain" and have designed a new computer chip "that uses the same wiring rules and architecture", resulting in a "flexible, scalable chip" that "operated efficiently in real time, while using very little power."

IBM has info on its site about the True North chip, also dubbed a "SyNAPSE" chip, with researchers calling it a "cognitive, brain-inspired chip to transform mobility and Internet of Things through sensory perception."

Click to enlarge
IBM's graphic on how the chip works.

Introduced on August 7, 2014, IBM says the chip "has the potential to transform mobility by spurring innovation around an entirely new class of applications with sensory capabilities at incredibly low power levels" through its "revolutionary new technology design inspired by the human brain". 

The chip is "powered by an unprecedented 1 million neurons and 256 million synapses", being the largest chip IBM has ever built with 5.4 billion transistors and an on-chip network of 4,096 neurosynaptic cores consuming only 70 milliwatts of power during "real time operation", which is vastly less power than traditional CPUs. 

IBM says that "as part of a complete cognitive hardware and software ecosystem, this technology opens new computing frontiers for distributed sensor and supercomputing applications."

Its researchers also say this particular architecture is "well suited to many applications that use complex neural networks in real time, for example, multiobject detection and classification", which will certainly be handy for future robots that we expect to interact as naturally with us as we do with other people. 

This detection was done with "400-pixel-by-240-pixel video input at 30 frames per second", and only consumed 63 milliwatts while doing so - which as the New York Times points out is less power than used by a hearing aid. 

Click to enlarge
What IBM's "True North" chip looks like inside.

The New York Times notes that traditional processors from Intel "only" have 1.4 billion transistors compared to True North's 5.4 billion, while using 35 to 140 watts, which is a far higher amount of power usage. 

The Times also quotes several people including Yann LeCun, a director of artificial interring at Facebook and a neural network pioneer suggesting the chip is still very limited and not that impressive yet, while also quoting Terry Sejnowski of the Salk Institute Computational Neurobiology Laboratory stating that "The True North chip is like the first transistor. It will take many generations before it can compete".

So, while we still don't have a chip that can power a robot android with a brain that can in any way, shape or form rival the capabilities of human brains, changing chips to work the way real brains do has been successfully achieved. 

Whether a version of the same chip in the distant future will ever achieve consciousness is yet to be seen, but humanity's journey towards the "singularity" where technology becomes conscious and smarter than humans has taken yet another step forward to reality. 

Click to enlarge
Will an IBM True North chip ever power one of these things?

Let's just hope we program in Isaac Asimov's famous three laws of robotics so one of these robot chip brains never ends up powering a Terminator.  

Here's another infographic from IBM with more detail. 

Click to enlarge
How the new chip has evolved since a prototype was developed in 2011.