Pages

Thursday, June 28, 2007

IBM supercomputer: Next step HAL?

Technology giant IBM has created the "fastest commercial" supercomputer capable of performing an astronomical number of calculations in one second. That kind of supercomputing power will allow scientists to carry on some pretty mind boggling work, and in all likelihood lead to some important technological advances.

Whether the real-world application of this work will be to the greater benefit is an open question that will depend on who is in control of the technology. Unfortunately, under the current social and political conditions, it's hard to be optimistic.

IBM's machine outstrips another supercomputer just debuted by Sun Microsystems. Here's an excerpt from the Guardian Unlimited's coverage:

Blue Gene/P is expected to be almost three times more powerful than its predecessor, and will run continuously at speeds of around 1 petaflop - one quadrillion calculations a second. It is also claimed to be more energy efficient than its rivals.
(see Scientists battle to build biggest supercomputer for the full story)

It brings to mind Stanley Kubrik's 2001: A Space Odyssey and its existentially challenged computer run amok. As you may recall, that computer was named HAL as an inside joke referring to the very same IBM, being a sequence one letter removed from the corporate moniker.

The more powerful our computers become and the wider their reach grows, the closer we get to realizing the kind of artificial intelligence that science fiction writers have depicted over the past few decades. In many cases those tale tellers have imagined worst case scenarios in which the technology comes back to haunt mankind, fulfilling the tragically necessary repayment of our hubris at mimicking the role of supreme being.

Getting beyond such doom and gloom, I instead like to ponder the ontological consequences of a truly artifical human intelligence. One possibility is that such a development could lead to the obsolescence of our very own mental powers, as we rely on computers to handle the intellectual chores necessary to take our civilization further, relegating ourselves to a simulated environment where our minds can float off into nothingness.

Yet again, maybe it will lead to a new evolutionary step for the human race, expanding the breadth of our own consciousness, connecting us all together in a way that we can harness our limited individual perspectives to achieve a greater, fully faceted awareness.

2 comments:

Anonymous said...

Artificial Intelligence is not a matter of Computing Power you got - its an architecture unknown to humanity. There's no computer architecture known that can "think" - why should a computer think? Even if its possible - he wouldn't have to. Humans think to survive (find food ...) - why should computers do that?

Francis Scudellari said...

Hi. Thanks for the comment. You're right that AI isn't directly related to computing power, but the more information a computer can process the more it can mimick the astounding powers of our own brain to solve problems and learn from past experience.

Why would computers need to think? Because humans want them to! The potential benefits from artificial intelligence are immense. And we humans think to do more than survive, we think to create and invent. What greater invention could there be than a machine that mimicks ourselves -- I know it's self flattering.

As scientist develop sensors and computers become more and more networked you'll see that artificial intelligence will develop, and it in fact may be arising in places you don't realize. Here are a few links to Wired Magazine articles, but there's much more available on AI and robots on that web site and elsewhere:

http://www.wired.com/wired/archive/10.03/everywhere.html

http://www.wired.com/wired/archive/2.04/cyc-o.html

http://www.wired.com/wired/archive/14.09/nextrobotics.html

http://www.wired.com/wired/archive/15.04/truman.html