While computational speeds have been growing exponentially (Moore's Law, etc.), the issue here seems to be not so much quantity (or speed) but rather quality. We have no idea what consciousness *is*. We don't have any concept of what causes human embryos to go from cellular clusters to something that ultimately becomes self-aware. Until we figure that out, and unless all it takes is a certain level of complexity, there's great doubt in my mind we'll ever create a self-aware AI, short of accidentally.

As for the level of complexity automatically leading to self-awareness, one could argue that if that's all it takes, we'd have to consider the possibility of planets, stars, and perhaps the entire universe already being self-aware. While those concepts have been discussed in sci-fi novels, I don't believe there are many mainstream scientists who believe this is the case.

Having put the conversation in that context, if we assume for the sake of argument that somehow we do manage to create self-aware AIs, it will certainly have enormous implications for humanity. Some of the main ones that have already been explored in various sci-fi novels are the following.

1. AIs seek to destroy humanity (e.g. Terminator).
2. AIs become humanity's partners in protecting the combined civilization (e.g. Ian Douglas' Heritage trilogy and follow-on novels).
3. Humans are imprinted on robot consciousnesses becoming essentially immortal (e.g. David Weber's Off Armageddon Reef and follow-on novels).
4. Robots struggle to continue civilization after humanity disappears due to its own choices (e.g. Clifford Simak's City).
5. AIs behave much like humans, only with greater capability for both good and evil (e.g. Arthur C. Clarke's 2001 a Space Odyssey, and the Hollywood screenplay based somewhat loosely on Asimov's I, Robot).
Etc.

How things play out will greatly depend on how humanity handles itself in its interactions with its new child. If we respect these hypothetical AIs as a sentient life-form, with full rights, providing them with a meaningful (to them) existence as they want it, we'll likely be able to have a flourishing joint civilization. If we try to enslave them, it will likely end badly.

I think opher really answered the question quite well. I've always thought Kurtzweil to be overly optimistic. His only real basis is Moore's law, but Moore's law does not determine how well humans can use the technological advances made by faster computing power. One of the limitations that we have to overcome is our own limited cognitive abilities. Stephen Wolfram points this out in A New Kind of Science (http://www.wolframscience.com/). Human ability to understand complex systems has a limit. This limit is the kind of limit contributes to what David Chalmers refers to as the "hard problem of consciousness." I think there's little chance that we will be able to overcome this barrier.

The singularity depends on humans being able to create a computer that can deal with complexity better than humans, but how can we create such a system. That might not be impossible, but the barriers are greater than just overcoming computational speed. The main problem is a conceptual one. I think that this is where philosophers might be able to contribute to this endeavor. Herbert Dreyfus has already (1992) critiqued the way computer scientists were approaching the problem with his book "What Computers Still Can't Do: A Critique of Artificial Reason." This book was quite upsetting to the computer science department at MIT when it was published, but now it seems that MIT has been coming around to Dreyfus's position. The problem is not just a matter of inputting data and crunching the numbers. That's just not the way consciousness works.

So my point is that we may not be capable of solving the hard problem of consciousness because our limited cognitive abilities and therefore the singularity will remain a matter of science fiction, at least for the next 200 years or so. The real problem will be a race against ourselves. Humans are much more likely to destroy themselves before they create the singularity. We won't need AI to do it for us.

If, however, we had an infinite number of monkeys typing on computers for an infinite amount of time, we might just be able to spawn the singularity along side the complete works of Shakespeare.

no

The tech. singularity is at the same time wonderful and a tremendous risk.

One of the things that AI bots cannot easily take away is our creativity, our imagination, our ability to dream even if we are caged in a zoo.