BackBy Brian Summerfield — April 2008
It’s nothing new for technological advances to be inspired by observations of the natural world. For example, Wilbur Wright — of the aviation pioneering Wright brothers — spent significant amounts of time looking at flying birds and noticed they sometimes “tipped” their wings to one side or another to gain balance and adjust to the differences in the lifting forces caused by the air around them. Unlike other aspiring aviators, he realized early on that the problem wasn’t one of powering a craft into the air, but of novel concepts such as lift and drag that helped sustain flight. As a result, he and his brother were the ones who made history as the makers and pilots of the first planes.
Another technology innovator has been in the news lately because of a cutting-edge project he’s working on that’s inspired by nodal-processing patterns within the human brain, or more precisely, the neocortex region. Jeff Hawkins — best known as the genius behind the PalmPilot — has spent the past few years of his career studying how these patterns can be applied to software design. He even formed a company, Numenta, to develop and promote the concept.
Specifically, this idea is manifested in the hierarchical temporal memory (HTM) model that ostensibly can be “trained” not only to recognize objects, but also identify and classify related objects it wasn’t trained on. The HTM system runs on the free-software Numenta Platform for Intelligent Computing (NuPIC) that was recently made available for download on the company’s Web site (http://www.numenta.com/for-developers/education/getting-started-htm.php).
Evidently taking architect Daniel Burnham’s advice to “make no little plans,” Hawkins wrote an introductory note to coincide with the launch, which included the following:
“We believe that the potential applications for HTM are broad and far-reaching; yet we know that our small company can only work on a piece of this application potential. Consequently, our goal is to create an industry built on HTM, opening up the platform to the world of creative developers.”
Now, this technology still is in its early stages — Hawkins actually termed it a “research release” — but clearly the implications of software that progressively learns the way a human brain does are staggering. This could have a huge impact on technologies ranging from databases to robots. Video-game maker Electronic Arts, for one, understands the stakes: The company is holding an HTM competition among game programmers, with the ultimate goal being to provide video-game characters that can learn and evolve.
Hawkins’ idea is not without its critics, though. In fact, one of them includes computer scientist and Numenta Co-Founder Dileep George, who said much of the work lacks a solid algorithmic basis. Still, he concedes that Hawkins’ hunches have taken the company in the right direction so far.
More disturbing is the idea that software will be able to evolve and one day learn to think for itself and not follow orders. This has been the plotline of many works of science fiction, including but not limited to the Terminator and Matrix series. If machines one day become our masters — and I’m not saying they will — we just might have Hawkins to thank for it.