For years, International Business Machines (IBM 1.22%) has been dabbling with what it calls "cognitive computing." Now the company that brought you the Watson supercomputer believes it has a chip that can think like the human brain.
Source: IBM.
Called TrueNorth, the chip draws on some 5.4 billion interconnected transistors to form a vast network not unlike the neural networks found in the human brain. That's a potentially massive breakthrough, especially for Internet-connected mobile devices that encounter new data every second.
A "right brain" computer
We're likely to be years away from mass production of the TrueNorth chip. And even then, experts quoted in this article in The New York Times seem to be split on its potential impact.
For its part, IBM isn't sparing the hyperbole. "The chip consumes merely 70 milliwatts, and is capable of 46 billion synaptic operations per second, per watt -- literally a synaptic supercomputer in your palm," IBM Fellow Dharmendra Modha wrote in a blog post describing the history and functions of TrueNorth.
It's a fascinating read. Modha says IBM started working on what became TrueNorth six years ago in an effort to create what he called a "brain-inspired" chip that would not only process vast amounts of data quickly but also detect patterns in the deluge.
Pattern recognition is what sets the brain apart. As Modha puts it: "If we think of today's von Neumann (i.e., traditional) computers as akin to the "left-brain" -- fast, symbolic, number-crunching calculators, then TrueNorth can be likened to the "right-brain" -- slow, sensory, pattern recognizing machines."
Think about that for a minute. With TrueNorth, IBM isn't going for speed or processing horsepower, but adaptability. Big Blue wants to build machines that learn.
IBM's history of cognitive computing
Watson isn't that kind of system. Think of it as the computing equivalent of a force of nature, blessed with so much memory and processing horsepower as to be capable of storing and recalling a near-infinite number of facts and combinations -- just the sort of machine you'd want if your aim was to win Jeopardy! or outperform a chessmaster.
As impressive as Watson is, Modha and his team at IBM have spent years working on a successor. In 2011, they unveiled experimental "neurosynaptic" chips as part of a project partly funded by the Defense Advanced Research Projects Agency, or DARPA. They called it a cognitive computing system:
Systems built with these chips won't be programmed the same way traditional computers are today. Rather, cognitive computers are expected to learn through experiences, find correlations, create hypotheses, and remember -- and learn from -- the outcomes, mimicking the brains structural and synaptic plasticity.
TrueNorth is the first realization of that vision. How long will it take to reach deployment? That depends on your definition of the word "deployment." For me, TrueNorth's true potential can't be tested till we see it in mobile devices.
Doubling your brainpower
And that's not just because of its low-power footprint -- though it has one -- but because connected smartphones and tablets encounter ever-shifting amounts of new data. TrueNorth's adaptability should allow for creative output in those situations, effectively doubling your brainpower.
Say you're at a meeting that's run 30 minutes long. Your TrueNorth-powered phone, deducing the same by matching GPS data with the current time, might pre-arrange (or pre-send!) a text to your next meeting's organizer notifying them of the delay. And that's a simple example.
TrueNorth's cognitive capabilities are also meant to discern context from multidimensional sensory data: light, sound, smell, and so on. That's of huge potential value in a mobile device operating in the real world, allowing for better, faster decisions on the ground. Either way, IBM's efforts could geometrically improve our ability to process information.
What might that mean for investors? With Gartner estimating that 2.2 billion smartphones and tablets will reach market in 2015 -- and perhaps much more than that in future years -- the revenue opportunity for leading this paradigm appears to be well into the billions. (IBM's Systems and Technology Group produced $5.1 billion in 2013 revenue.)
It'll take years of hard work and luck to cash in, of course. But if anyone can, given what we already know about TrueNorth, it's probably Big Blue.
