One Small Step for Man, One Giant Leap for AI

In the 1960’s a space race style competition was going on internationally with the ultimate realization when man first actually made it to the moon and set foot on its surface.  After this, the cold war ended shortly afterward and as a result competitions like these have seemed lest urgent.  Instead, the profit driven world of more terrestrial matters has supplanted the “less profitable” ventures such as space exploration.  But there may soon be an upcoming race that will be both as profitable and as promising to mankind as the space race: the race of super computing.  And it may have just as much of an effect on our lives as the race into the stars – without the same limitations.

AI has long been advocated by theorists suggesting that the coming singularity and fusion between man and machine could one day rocket us into a world strange and alien to us and yet just as full of promise as the stories where humans may end up in space one day.  But putting a human being on the moon and creating the ultimate computer processing facility could have far different results.  But who are the two top competitors in this race toward an artificially intelligent computer?  And what hurdles would we have to overcome to create a truly intelligent machine?

In 2008, the Road Runner Computer at Los Alamos National Laboratory in New Mexico was considered the world’s largest computer.  The system was capable of performing 1.026 Petaflops.  Each petaflop is approximately 10 to the power of sixteen floating point operations per second.  Once thought to be more processing power than could be processed on the entire surface of Earth, a petaflop is only the latest generation of calculations with many more expected to arrive in the future.  Then, after the Road Runner there was the Cray Jaguar supercomputer in 2009 which processed at a rate of 1.759 Petaflops per second.  Several computers later, in 2011 the Fujitsu K Computer was created with a processing power of 10.51 Petaflops.  Rather than slowing down, the processing speed of these devices has been increasing significantly in recent years with no sign of slowing down.  Although there is one major difference in how this compares to the space race.  While the space race was an important issue of nationality, this will prove to be an important issue of corporate identity.

But what limitations will there be that might keep us from seeing the first sentient computers?  Some have argued that the IBM Roadrunner had within its 1.026 Petaflops the requirements to make the an AI fully, but that the issue may be one of software not hardware.  It’s not the size of our brains that make us what we are, in other words, but the way in which it is arranged.  It’s not a hardware issue, but a software one.  And so far a system for programming artificial intelligence to be humanlike has not been discovered.  It would perhaps be simpler to just evolve the system from scratch so it will go through a process similar to the one humanity did.  Of course even that would have its problems as it would be developing in a digital reality completely different from our own.