Neurogrid: brain-modeled microchips

It just sounds futuristic.

Neurogrid.

Could have popped right off the page of a cyberpunk book.  But it's a reality.

It's a microchip developed at Stanford and modeled on the workings of the human brain.  It also happens to be 9,000 times faster than chips found in the average PC.  As said previously, this innovative design has as much to do with neurology as engineering.  The human brain can still outdo most machines as George Dvorsky nicely summarizes in the io9 article:

"It's [the brain] a highly efficient information processor capable of crunching 100 million instructions per second (MIPS). Astoundingly, it only uses about 20 watts to power its 100 billion neurons. Today, our best supercomputers require a million watts to simulate a million neurons in real time (measured in terraflops). A standard desktop computer requires about 40,000 times more power to run and operates about 9,000 times slower."

So the idea is to create a chip that approaches this kind of crunching while still being energy efficient.  Neurogrid is an impressive step towards that goal.  An arrangement of Neurogrids can simulate one million neurons while only consuming three watts of power.  All this and a deeper understanding of how the human brain works.  Win-flippin'-win, baby.

Many of the end benefits of this technology seem to sit squarely within robotics and cybernetics.  A robot with a Neurogrid core would be more autonomous in terms of power needs (cue irrational fears of killer robots with aimless wanderlust.)  Neurogrids would also help make quantum leaps in cybernetic limbs, allowing for real-time transmission of signals between the brain and the appendage and thus more fluid movements.  My thoughts turn to something else, however.

Ray Kurzweil has taken knocks for his goal of uploading a human mind to a computer.  Part of the criticism has been that there is just entirely too much information to process in order to accomplish such a feat.  Devices such as Neurogrid, ones that promise to handle nigh inordinate amounts of data at little cost of energy, might be how to get around that problem.  One hundred years from now, what will processing speeds be like?

I'm still holding out for either a cybernetic body or be fully transcendent before that century mark, though.





Follow me on Twitter: @Jntweets