Singularity is Still a Ways Away

Transistor_Count_and_Moore's_Law_-_2011.svg (1)Computer Science pioneer John Von Neumann, who designed the architecture that characterizes nearly every computer in use today, long ago coined the term singularity to describe the moment in time when artificial intelligence would exceed the power of human intelligence. At the point of singularity, the capacity of computers to think would exceed the intellectual capacity of mankind, machines could then create even smarter ones, and the course of humanity would become unpredictable at best, apocalyptic at worse. Various predictions forecast that singularity is 80% likely to occur between 2017 and 2112. Famous futurist Ray Kurzweil predicts it will occur before 2045.

We humans can take solace from this article, however, which describes just how puny computers’ intellect currently is compared to humans’. Using all 83,000 processors of the fourth fastest computer in the world, the Fujitsu K in Japan, computer scientists created a simulation of 1.73 billion brain cells connected by 10.4 trillion synapses. Each synapse was modeled using 24 bytes, thus consuming a total of 1 petabyte of memory. That’s 1 quadrillion bytes, which is equivalent to the memory of 260,000 of today’s typical laptops. Even with all that memory and horsepower, however, the simulation captures only a tiny fraction of the brain’s full structure. The human brain consists 200 billion brain cells and hundreds of trillions of synapses. In other words, one of today’s most powerful computers is able to simulate only about 1/100th of the human brain. In terms of performance, the K machine was able to emulate one percent of one second’s worth of the brain’s activity, and even that took 40 minutes.

Moore’s Law predicts that computer performance doubles every 18 months. That trend has held true for over forty years now. So, computer performance can be expected to grow by a factor of 10 in around 5 years, and by a factor of 1,000 in 15 years. That kind of improvement would seem to put us within striking distance of singularity in the year 2028. That seems like a long way off, and Moore’s Law may start to hit limits before then due to physical constraints that have long been predicted but have never actually come to fruition.

For now, though, it looks like computers will continue to do our bidding for a little while longer – except, of course, for iTunes. If anyone has figured out how to get that software ode to Rube Goldberg to do your bidding, you have my admiration.

 

Ray Klump

About Ray Klump

Professor and chair of Mathematics and Computer Science Director, Master of Science in Information Security Lewis University http://online.lewisu.edu/ms-information-security.asp, http://online.lewisu.edu/resource/engineering-technology/articles.asp, http://cs.lewisu.edu. You can find him on Google+.

Leave a Reply

Your email address will not be published. Required fields are marked *