From the IBM Research blog:
4. These systems won’t be programmed like traditional computers are today. Rather, cognitive computers will learn dynamically through experiences, find correlations, create hypotheses and remember – and learn from – the outcomes, emulating the human brain’s synaptic and structural plasticity (or the brain’s ability to re-wire itself over time as it learns and responds to experiences and interactions with its environment.)
5. To accomplish this new kind of system, IBM is combining neuroscience, nanoscience and supercomputing together to rival the function, power and space of the brain.
I find this stuff both incredibly fascinating and somewhat scary. With all the science-fiction literature we’ve been exposed so far, it’s hard not to imagine disquieting scenarios where the computers start being completely self-sufficient and take over the world. (But who knows, maybe they’ll manage to do a better job than us!)
See also The official IBM press release page.