There’s an interesting article in the Technology Review by MIT, written by Paul Allen regarding the Singularity concept. The gist of the article is that the Singularity is, in fact, not so near because some of the breakthroughs required to bring machines up to super human level intelligence do not follow an exponential growth rate. I think this is a fair criticism, especially regarding replicating consciousness. Although, Ray Kurzweil (creator of the Singularity concept) does address the software issue in his book (basically it lags slightly behind hardware and if memory serves, was slightly harder to predict). However, I also remember reading this article and being surprised (basically, software performance has accelerated faster than hardware in the recent past).
It seems to be at odds with what Paul Allen is saying to some extent. Paul Allen’s best argument here is that (1) there are fundamental software programming “walls of understanding” (like understanding exactly how the human brain creates consciousness) that we need to break through and that (2) solving these problems will not follow that exponential curve of progress. It’s not necessarily a matter of increasing software performance (that part seems on track, as does hardware). It’s a matter of understanding how to create a thinking, conscious machine that is capable of being as creative as humans.
Paul Allen also mentions a bird. I have heard Daniel Dennett (who also does not think the Singularity is near) bring up the same “recreating the bird exactly how it works in nature” argument. Basically it goes like this… We could build the exact replica bird in principle (as Dennett said, it would be an effort that dwarfs the Manhattan project), but why would we spend resources in this manner? Why not just borrow some of the qualities of the bird for our technology that are much easier to recreate as a function – feathers, flight, etc. and we get to the same end, but not in the same way.
Personally, I’d rather augment my own abilities á la Tony Stark heads up display to do computational and analytic tasks (people are good at generalizing – think of all the calculations that have to go into catching a ball, not math – I struggle with mental long division and that’s just a single calculation).
Bottom line for me – If I can augment my own computation abilities in increasingly convenient ways (right now I have to tote around this laptop- I’d rather have a cool heads up display on my glasses that I control with my brain that leverages the power of the internets/cloud/whatever), then I bet I could be much more productive (or maybe slightly more productive, but in a 1/5 of the time). I also bet scientists working on AI could be more productive too. Clearly, progress will march on. While the Singularity is possible in principle, I am skeptical of the 2045 date. I am however hopeful of doing some other super cool stuff, even if computers are still not conscious entities with mental faculties many times that of humans.
Another resource: I found this pdf to be very informative for the who’s who of opinions regarding the Singularity.