Ok, this is an answer about AI. But obviously technology won't stand still, and is moving very quickly on multiple fronts. So part of the question was, whatever you think it going to happen, how will the world look in 20 or 30 years?

OK but that was actually not what you asked. You asked, "Do you think "The Singularity" described by Ray Kurzweil will happen?" The Singularity is the (theoretical) point at which machine computational power will catch up to human brain computational power. Kurzweil thinks that when that happens, machines will be able to think, and this the 'Age of Spiritual Machines' will be upon us.

So I obviously don't think that will happen. Machines will simply continue to be mirrors of ourselves, not standalone entities.

In 20 or 30 years I think things will look fairly similar to how they are today. We'll have some surprises, of course, but if you look at the last 20-30 years and the predictions of the 'future' we live in now, they are pretty wildly optimistic: 2001 a space odyssey had us with bases on the Moon and manned trips to Jupiter; 'Things to Come' in the beginning of the 20th century predicted a future filled with technology we are still far from achieving. In short: there are no flying cars. Today is fairly similar to the 1970's or 80's. On the other hand, we have the Internet and iPhones and iPods. That is very different and surprising. The Internet was an utter surprise even to Bill Gates -- recall that Microsoft slept in ignorance for a full year and half while Netscape spread like a weed (and to their credit, the war machine woke up, built IE, attacked and won).

But the point is: the Internet has changed fundamentals of social interaction more than anything else, imho. Today you see groups of kids, all huddled in a circle, oblivious to one another as they all text other friends far away.

So: 20-30 years? Pretty much like now, but with some discontinuous innovations like the Internet that are an utter surprise to everyone, with ramifications that are more social than anything else.

While I agree the answer is no, I do not agree with your arguments.

That all attempts to date are failures implies future failures is a fallacy. If true, no form of innovation is possible.

While you are correct that Kurzweil's supposition is flawed, it is not proof that no such supposition is true.

While computers do follow Von Neumann state machine architecture, they need not be the deterministic machines you describe. Probabilistic approaches are (arguably) the forefront of modern AI. Further, it is reasonable to assume that new hardware approaches (for example, auto associative memory) may be introduced into computers in the future. If we assume such conventions are a prerequisite for AI (i.e. current computing machines are not capable), we can also say that it is reasonable to believe we could construct such an architecture.