Kursweil's claims, although possible, are not supported by evidence and frequently rely on an oversimplification of intellience, consciousness, and human ability. I, too, once believed his claims for the future. Then I challenged his premises.
I tend to agree but is he doing so to explain the complexities in simplistic terms so the masses would "get it". Jesus the intellectual spoke in parables so the masses would understand. Confucius was no different. And let's not forget Albert's words of keeping it simple.......but not to simple.
Kurzweil is respected as a person in the AI community, but most core research theorists find his grasp of some fairly fundamental issues to be lacking. He makes a good pop-sci figurehead for more mundane aspects, but his understanding of the fundamental theoretical nature of the problem is quite poor. His theory is stuck in a 1980s era model, which we know is broken.
These models are like religion for many researchers, and once they latch on they never let go. Part of the problem is that there are many theoretical paths to intelligent implementations, but very few tractable paths. Yet once a path is proved to be intractable, many researchers continue on that path, neither addressing the intractability nor switching to a better model. Human nature I guess.