September 27, 2005

Ray Kurzweil & the "Singularity"

Ray Kurzweil has a hypothesis. He thinks that technological innovation is increasing so fast that, in a relatively short time, it will change humanity so completely that it is impossible to predict what human existence will be like in 20-50 years (or so). The convergence of information science, computing, biotech, and nanotech will cause a fundamental shift in the pace of technological expansion..so huge a shift that he has given it a name: the "Singularity".

Ray wrote a book on it. Or three. His latest one is currently #22 in book sales at amazon. The idea is popular. Very popular. Why do I mention it here? People who believe in this idea (they're called "Singularitarians" or something) really love the idea of synthetic biology because they see it as the next stepping stone towards bionic ... everything.

I'm not going to deny the appeal of the idea or try to dissuade people from believing in it. It may very well come true, eventually. However, Ray describes technological innovation as an exponential process. (As time increases, the technological state of the art increases like exp(t) ). I also think that technological innovation is exponentially increasing.

But there's a funny thing about the exponential process, exp(t) ... or exp(r*t) if you want to include a rate of growth. If you start from an initial time and watch the growth of technology, it looks grand and awe-inspiring. But if every few time periods (years? months?) you get used the state of technological innovation, then the rate of technology growth no longer feels exponentially increasing. Instead, if you look back in time you will see more or less linear growth and if you look ahead in time you should expect linear growth. Why?

Well, if you split up exp(r*t) into the growth you got used at time to, which is exp(r*to), and the growth you should expect in the future, which is exp(r*(t - to)), then you get exp(r*t) = exp(r*to) * exp(r*(t - to)).

If you Taylor expand this and consider (t - to) to be small (ie. the near future) then you will get: exp(r*t) = exp(r*to) * (1 + (t-to) ).

What does that mean? If you already got used to computers, GM foods, iPods that can fit in your jeans pocket, and just the idea of a Space Elevator, then you should NOT expect miraculous technological innovation in the near future. And if you continuously get used to the innovation that regularly occurs in the marketplace and in academia than you will never be completely surprised by what's coming around the corner.

So, the people of the future will never experience a Singularity. If the people of the past were to somehow catapult themselves into the future, then they will experience the Singularity. But that holds true if they are cavemen experiencing Medieval Europe or Renaissance Europe experiencing modern tech.

I think the future will still hold amazing feats of technological innovation. But, by the time they occur, we will be more or less ready for them. At least, we will be more ready for them than the word 'Singularity' implies. I think anyone that suggests future tech will somehow supplant humanity's human-ness needs to read more Shakespeare.

Posted by sali0090 at September 27, 2005 9:59 PM