AI Researchers Build Brain: Singularity Watch?

Posted by Dr. James Canton on April 1, 2009 under Uncategorized | Be the First to Comment

This article comes from MIT’s Tech Review and is startling in its implications for what is next in not just computing, but in driving the Singularity Forecast. Will we witness an era where AI rivals human intelligence or even surpasses it?

Perhaps a different form of AI, different form of consciousness from human beings, a synthetic form of intelligence I would maintain, has already emerged.

I wrote about this in my article When the Network Wakes Up. In it, I posed the notion that perhaps the Singularity has already started and we did not notice because we are so human-centric in looking for consciousness and intelligence that is biological, or computational in form, when there is a new A-Life model that is emerging, a mash-up of silicon and biology. Read on.

Artificial intelligence investigators have built a fully silicon scale simulation of the human brain. The artificial neurons operate faster than the organic model, are built to learn and adapt.

The Fast Analog Computing with Emergent Transient States project takes a different approach to other electronic intellect endeavors. Research like the Blue Brain project, run vast software simulations of virtual brains, which allows them to tinker with the conditions and wiring of the brain with the tap of a keyboard. On the downside, you’re running a layer of simulation of a parallel system on top of an utterly sequential computer system, which slows things down.

The FACETS hardware instead builds direct silicon similes of synapses and neural circuits, creating a real hardware brain which can operate in parallel just like the human mind. Sure, it’s more of an American Idol mind at the moment, with only two hundred thousand neurons compared to a hundred billion in your head (a factor of five hundred thousand).

But the FACETS architecture is scalable, and the team already has plans for a billion-synapse super chip – for those of you updating your “end of the human race” calendars, that’s a fifty-thousand-fold increase in one generation. And we all know that computers don’t have new generations more than once or twice a year.

FACETS Artificial Brain