The brain is an amazing thing. "It consumes less power than a light bulb and occupies less space than a two-liter bottle of soda," writes Dharmendra S. Modha, manager of cognitive computing at IBM (IBM), in his blog. Yet it performs functions no computer can -- many of them quite basic. Think of those blurry, squiggly jumbles of letters and numbers that web sites use to make sure they're dealing with a human being. Your brain can sense, perceive, reason, and coordinate different functions in a constantly changing environment; it can handle ambiguity and abstraction.
There's not an app for that -- yet. But computer scientists at IBM have a longterm goal of achieving cognitive computing, a mindlike artificial intelligence, allowing computers to handle far more complex systems, such as finding patterns throughout the Web, the way humans can pick out a face from a crowd.
Thinking Like a Cat
To that end, Big Blue announced two milestones last week: simulating a cat's cerebral cortex and unveiling a new software that maps how a brain works. For the first, scientists and university collaborators used the Blue Gene supercomputer (with 147,456 processors and 144 terabytes of memory). The simulation ran 100 to 1,000 times slower than a real cat's brain, but it simulated one billion neurons and 10 trillion synapses -- more than the amount in a cat's brain.
The simulated cat's brain isn't a virtual cat. The simulation is intended as a tool allowing researchers to study behavior and dynamics within the brain.
The scientists also unveiled a software application called BlueMatter. Using a certain type of MRI, the software can map the workings and connections of the brain's regions and how they communicate when given stimuli.
Obviously, an array of supercomputers that still can't do basic brain functions isn't close to being an artificial brain. But as researchers make advances in neuroscience and computer science, they're closer to building a computer that can emulate a brain's function. IBM's goal is "building a compact, low-power, compact cognitive computers that approach mammalian-scale intelligence and use significantly less energy than today's computing systems."
Up Next: Monkey Business
Cognitive computing could be used in numerous applications: to monitor complex systems such as weather and traffic, to find and analyze patterns in large amounts of data, and to predict problems while accounting for context and previous experience. IBM predicts these innovations will spark new industries, and the Defense Advanced Research Projects Agency is interested.
Next, the researchers plan to simulate a monkey's brain. Simulation of a human cortex could come within the next decade, assuming a continuing rate of advances in computer chips and memory storage, Modha says. A human scale cortical simulation will probably require four petabytes of memory -- that's 4,000 terabytes, or 4 million gigabytes -- and running these simulations in real time will have to perform more than one exaflop/s, or 15 quintillion calculations per second.
Soul of a New Machine
But simulating the brain's brawn is the relatively easy part. The real challenge will be achieving the brain's essence: how to wire up a cognitive system. Neuroscience, despite significant advances, is still far off from achieving that. So it's far too early to estimate what this feat could mean to IBM, but we can expect great things. It's in the company's heritage -- maybe in its DNA -- to stay at the forefront of technology and computer science.
For now, other questions arise, many of them seemingly from the realm of science fiction. Could a cognitive computer gain consciousness? Could it be self-aware? And how would that change the meaning of the concept of the human soul?
Could human brains be uploaded to computers and become entities run on a machine or a robot? And if cognitive computers could mimic all brain functions, they could also learn -- so could they then progress and improve their own designs beyond human intelligence, thus achieving the concept of singularity? For now, we'll have to wait and wonder.
Why do investors make the decisions that they do?View Course »