The Singularity is Coming, or Is It?

There has been a lot of buzz about a new book by Ray Kurzweil predicting "The Singularity". Here is one example. Here is another. This one goes into detail.

Basically, when the Singularity comes, humanity will merge with self-aware computers to become god-like. Normal humans will find themselves obsolete.

I'm not worried for a couple of reasons.

First, Kurzweil is predicting based on current trends. One of these is Moore's Law which says that the complexity of computer chips doubles every 18 months. Kurzweil uses this to predict how complex computers will be in 40 years.

The thing is, Moore's Law is actually Moore's Observation. Moore, a top engineer at HP at the time, observed in the 1970s how fast compuer complexity increased. This became self-fulfilling when the marketing people at Intel got ahold of it. Currenty Intel executives are convinced that their stock will drop precipitously if they fall behind Moore's Law so they throw enourmous resources at maintaining this rate. This means that rather than being a hard and fast law, it is a marketing concept.

Eventually Moore's Law will run into physics. To see what I mean, look at flight.

The Wright Brothers first flew in 1903. Their flight was nothing more than an extended hop. They realized this and kept it secret for nearly two years. By 1905 they had an airplane that could stay in the air for a half hour (until it ran out of fuel) and was fully controlable. It could fly at speeds of nearly 40 miles an hour.

Fifteen years later planes could fly reach speeds of over 100 miles per hour. By WWII jets were invented and the speed of sound was broken in 1947. The SR71 Blackhawk first flew in 1964. It could fly at mach 3.3 (three times the speed of sound).

So we went from 40 mph to 2,175 mph in sixty years. If we extropolate from there then we should have planes that can fly 10,000 mph or faster but the SR71 continues to be the fastest plane built. Flight ran into the laws of physics. Friction is a real problem at these speeds, both as drag and as heat.

This is happening with computers. Computer complexity is increased by making finer etching in silicon but we are approaching the limits. You cannot etch smaller than an atom.

Speed is another issue. Clock speeds keep increasing but there are limits here, also. We already hit the limits of memory. While engineers put more memory on a chip they have not made it ru any faster in some time. There are tricks for getting around thisbut they are tricks and they have their own limits.

So the idea of Moore's Law continuing for the next 40 years is unlikely.

The next problem with Kurzweil's predictions is self-aware computers. He assues that once computers are powerful enough they will be self-aware.There is no reason to think this. We think in ways that are very different from the way computers work. While scientists are researching exactly how we think, we do not know when this will be solved or how computers will need to be changed to simulate thought.

Kurzweil makes a lot of other predeictions dealing with genetics ond other sciences but I doubt that he is any closer with them.