IBM’s New Processors Will Think Like Brains

It’s been just over 30 years since IBM released its first PC and shook the whole world up. Hoping to do some more world-shaking, they’ve now built two chips that function more similarly to our brains than normal chips do.

Yes people, this could be the day we all look back upon with sheer horror, as The Robots chase us from our houses and onto the streets.

Funded in part by DARPA, which contributed $US41 million to the project, the prototype chips use parallel processing to multi-task, but more importantly adapt to new information even if they’re not expecting it. The projector leader at IBM Research, Dharmendra Modha, claims that “you have to throw out virtually everything we know about how these chips are designed”, elaborating that “the key, key, key difference really is the memory and the processor are very closely brought together. There’s a massive, massive amount of parallelism.”

They also have sensors for identifying changes in the real world, such as temperature rises, audio and motion movements, which could make them well-suited to outdoors use, in cars, traffic lights or planes for example.

While it’s unlikely we won’t see these two prototype chips put into practise for another 10 years, IBM is confident it’s the future for their business. Both chips have 256 neurons, with one of them containing 262,144 “programmable synapses”, and the other 65,536 “learning synapses”, which will be built upon and hopefully get to the point where it’s possible to have a chip with 10 billion neurons, and 100 trillion synapses.

It’s likely those numbers sound impossibly large to you — personally, I can’t even fathom the size and power. But IBM’s Modha doesn’t give any reason for us to be scared, telling the NY Times that they’re “not trying to build a brain”, that they’re just “trying [to]draw inspiration from the brain”. [Yahoo, WSJ and NY Times]