Human life: The next generation

New Scientist has an article entitled Human life: The next generation by Ray Kurzweil, which suggests that the rate of technological advance is such that it won't be long before we significantly upgrade humans to a better model.

The key to Kurzweil's argument is Kurzweil's Law (aka "the law of accelerating returns"), which says that future advances will give us an exponential growth of technology, rather than merely linear growth. Potentially, that could mean enormous advances over a human lifetime. Past experience shows that this may indeed be true.

The problem is to predict in what direction technological growth will occur. You know that the technology is going to be mind-bogglingly more advanced than current technology, but you don't know where these advances are going to manifest themselves.

Kurzweil says that:

"...information technologies will grow at an explosive rate. And information technology is the technology that we need to consider. Ultimately everything of value will become an information technology: our biology, our thoughts and thinking processes, manufacturing nd many other fields..."

I agree with all of that.

We are way past the stage where owning a steam engine was a good way of investing your money. Useful engines move bits nowadays.

Kurzweil also says that:

"...By the 2020s, nanotechnology will enable us to create almost any physical product we want from inexpensive materials, using information processes."

That is mostly rubbish.

It is certainly true that nanotechnology holds the potential for doing this. However, the design of a nanotechnological "factory" may require far more information than we can accumulate in the few years between now and 2020.

An example is manufacturing a human being, or a fish, or an ant or whatever. Assuming you start from nothing I don't think you will be able to manufacture any of these by 2020. We can already make a virus starting from nothing, but the process is rather hands-on.

Surely, Kurzweil is not suggesting that the same hands-on approach can be used for making more complicated objects? I assume not.

The production process has to be automated somehow. The obvious way is to first of all design a set of simple "tools", that then take over to do the next stage of designing more complex "tools", and so on up the scale of complexity, until you arrive at the object you wanted to manufacture in the first place. This sounds very much like the sort of solution that evolution discovered over a rather long period of random shuffling about and selection by the environment for "fitness".

How exactly is this evolutionary type of process going to be compressed into the time remaining between now and 2020? Er ... it's not. The reason is that Kurzweil assumes an exponential growth rate that is far too fast if his Kurzweil's Law is to achieve everything it needs to between now and 2020.

Nevertheless, Kurzweil's Law is probably broadly correct, and we will be able to drive evolution forwards at an increasing rate, so his future with nanotechnological "factories" making almost anything we want is much closer than we think it is. The trick will be not to specify in advance which particular complex objects should be made, but to wait and see which complex objects are feasible to make, and then to try to match these objects up with applications that we find useful.

Having said that, we'll still have some really cool technology in 2020, and some of it will be built by nanotechnological "factories".

Update: There is an interesting book called Accelerando that follows up the consequences of Kurzweil's Law. This was pointed out by John Baez here.