IBM recently unveiled a new “cognitive computing” chip that is said to function like the human brain. These sorts of claims are made all the time, and are, as a rule, grossly exaggerated or outright false. But from everything I’ve read, this breakthrough sounds legitimate: if the researchers have done what they claim, this may be a significant break from the integrated, highly linear Von Neumann architecture that has been at the heart of computer processors for over sixty years.
The reports I’ve read, though, have all mostly failed to emphasize the likely fact that this new architecture can only be taken full advantage of by tasks like pattern recognition that are already amenable to being processed in parallel, or all at once, rather than in a sequence of steps.
Also, reports of the brain’s demise are greatly exaggerated: this architecture is a lot more like the brain than current architecture, but it’s still not remotely similar enough to produce consciousness, thought, or strong AI.
UPDATE: See also Alex Knapp’s two highly informative posts on the IBM research: “Is IBM Building a Computer That Thinks Like a Human?” and “How IBM’s Cognitive Computer Works.”