Jun 25, 2016
Brain-like computers may now be realistic
Posted by Shailesh Prasad in categories: 3D printing, biotech/medical, computing, nanotechnology, robotics/AI, transportation
Power consumption is one of the biggest reasons why you haven’t seen a brain-like computer beyond the lab: the artificial synapses you’d need tend to draw much more power than the real thing. Thankfully, realistic energy use is no longer an unattainable dream. Researchers have built nanowire synapses that consume just 1.23 femtojoules of power — for reference, a real neuron uses 10 femtojoules. They achieve that extremely low demand by using a wrap of two organic materials to release and trap ions, much like real nerve fibers.
There’s a lot of work to be done before this is practical. The scientists want to shrink their nanowires down from 200 nanometers thick to a few dozen, and they’d need new 3D printing techniques to create structures that more closely imitate real brains. Nonetheless, the concept of computers with brain-level complexity is that much more realistic — the team tells Scientific American that it could see applications in everything from smarter robots and self-driving cars through to advanced medical diagnosis.