So are supercomputers really becoming more like human brains? Well, Japanese computer scientists have simulated cranial activity for one second, and now commercial chips are coming out that learn from their own mistakes, the New York Times reports. While today's computers are pre-programmed to perform operations with strings of 0s and 1s, the next generation of processors will function more like biological synapses. Instead of being programmed, circuits will be "weighted" by data they have "learned"—and new data will cause those values to change.
Designers say the new "neuromorphic processors" may lead to computers that safely walk and drive around—but the dystopian future of killer renegade robots is nowhere close, and even our current computers won't be replaced. The new circuits will likely augment current ones in the cloud and in smartphones, partly to prevent crashes. A sign of their popularity can be seen at Stanford University, where a machine-learning course has drawn more than 760 students. "That reflects the zeitgeist," says a neuroscientist. "Everyone knows there is something big happening, and they’re trying find out what it is." (More computer chip stories.)