Kurzweil's book tries to predict what our lives would be like in the year 2100 (yes, one of his predictions is that we'll all still be "alive" in 2100 - for the reason I put "alive" in quotes, you'll have to read the book).
A common theme you see in many science-fiction books and films that try to depict life on earth in 2100, or life of advanced aliens, is the striking similarity between the way of life of these creatures and our current lives. Star-Trek is a good example. Sure, Captain Kirk shoots a laser gun, gets teleported and eats food generated by a machine, but in his world humans (or other carbon-based life forms) still rule, travel physically in the universe, get cured by a human doctor, and so on. More unusual life forms are either relegated to one episode, or given bizarre flaws to explain their rarity (e.g., Commander Data).
So, what will earth really look like in 97 years, in 2100? What will it look like in just 17 years, in 2020? Kurzweil sets out to predict the answers to these questions, and he does so in an enjoyable writing style and using his extensive technical knowledge and visionary approach. He will shock most readers by his predictions which initially seem outlandish, but on second thought suddenly sound very reasonable and very possible - and perhaps even - undeniable.
The basic premise of this very interesting book is what Kurzweil calls "The Law of Accelerating Returns". Moore's law, stating (roughly) that the computing power of a $1000 computer doubles every 12 months, is an example of Kurzweil's more general law. But Moore law only talks about integrated circuits made from transistors - this law only became relevant in the 1960s, and will most likely stop being relevant sometime in the next decade. But Kurzweil demonstrates that the same "law" of computing acceleration has been valid ever since 1900 (!): The first computers were mechanical, then came computers using electro-mechanical relays, then came vacuum tubes, then stand-alone transistors and finally integrated circuits and VLSI; Computing continued to accelerate at an almost constant pace throughout all these changes in paradigms and technologies, and Kurzweil argues that it will continue to do so - even if we need to replace our IC-based computers by computers based on massively-parallel neural networks, nanotechnology-manufactured computers or even quantum computers.
Once you understand Kurzweil's basic premise and agree that it is plausible (he explains it very well and very convincingly), the unavoidable consequences are staggering. The most obvious thing that is going to happen if computing accelerates in its current pace, is that around 2020, a $1000 computer will have the computing power of a human brain. Very quickly afterwards the computer "intelligence" will surpass those of humans. In the following decades other advances in technology like self-replicating nanotechnology will make relying on human labor and thinking not only unnecessary - it will even be stupid. Sending a human for exploration missions in outer space in a large UFO-like spaceship would be extraordinarily silly, when you could send a computer sized like a grain of rice and having the intelligence of a thousand humans. By 2100, computer intelligence and the original human intelligence that started it all will be completely inseparable, according to Kurzweil. I don't want to spoil your fun of reading the book, so I won't reveal here more of Kurzweil's predictions.
Kurzweil's book isn't perfect, of course. It discusses philosophical and moral issues very sparingly. It downplays "modes of failure" (like computer viruses, renegade nonobots) and the effect of Luddites and underdeveloped countries. It is very conservative economically (Bill Gates will remain the richest person in 2050, in 2020 there will be many more lawyers than doctors, because Intellectual Property will be the most important economic issue).
All-in-all Kurzweil's book is very thought-provoking and I strongly recommend it. Even if most of his predictions never come true, it really shines a light on the question of what might happen as computers get stronger and stronger, too strong to be used merely as a platform for "cute" GUIs like Mac OS/X or MS-Windows :)
