IBM has announced a new chip that it says is a breakthrough in emulating the human brain:
"Power is the fundamental constraint as we move forward," says Horst Simon, deputy director of Lawrence Berkeley National Laboratory, a major supercomputer user. "This chip is an indication that we are really at the threshold of a fundamental change in architecture."
....TrueNorth, IBM says, uses 5.4 billion transistors—four times more than a typical PC processor—to yield the equivalent of one million neurons and 256 million synapses. They are organized into 4,096 structures called "neurosynaptic cores," each able to store, process and transmit data to any other using a communications scheme called a crossbar.
The design is "event-driven," Mr. Modha says. That means that individual cores fire up only when they are needed, rather than running all the time. This scheme makes the chips more power efficient. Where a comparable standard microprocessor draws 50 to 100 watts per square centimeter, TrueNorth draws just 20 milliwatts, or thousandths of a watt, IBM says.
That's about as many neurons as a small insect has. You'd need something on the order of 100,000 of these chips to provide as many neurons as the typical human brain—though that's probably not really a meaningful number. If digital neurons are faster than chemical neurons, you might need fewer of them. You also don't need any of the neurons that are designed solely to keep the body physically alive. And traditional chips can pick up a lot of the load too. On the other hand, the 3-D structure of the brain provides some advantages you don't get from a 2-D chip.
In other words, who knows? Maybe you need 10,000, maybe you need a million. Maybe this whole approach will turn out to be a dead end. And we're still a long way off from developing the software to make this all work in any case.
Still: it's cool stuff. There are lots of different approaches to developing artificial intelligence, and this is certainly a plausible one. It probably won't take too long before we know whether it really holds the promise that AI researchers hope it does.