I, Robot

While at MIT he developed a computer program that helped high-school kids choose the right college. He sold the program to a publisher for $100,000. Next he started a company that developed and integrated three technologies—optical-character-recognition software, a flatbed scanner and a text-to-speech synthesizer—to create a machine that could read documents to the blind. Stevie Wonder bought one, which led to a friendship with Kurzweil and to Kurzweil's next product, a music synthesizer that could re-create the sound of real acoustic instruments. Kurzweil sold the reading-machine company to Xerox in 1980, and sold the music-synthesizer company to Young Chang, a musical-instrument company in Korea, in 1990. Since then, Kurzweil has dabbled in technology for education and medicine. He now runs a hedge fund called FatKat, which uses artificial intelligence and pattern-recognition software to pick stocks, as well as Kurzweil Technologies, a 20-person outfit that invests in early-stage tech companies and incubates ideas of its own, too.

But even as he was building his companies, Kurzweil harbored a passion for artificial intelligence. He was consumed by the idea that computers might someday enable us to extend our lives, or perhaps even make us immortal. The seeds for this thinking lay in the loss of his father, Fredric, a composer and conductor who died of heart disease in 1970, when Kurzweil was 22 years old. "I find death unacceptable," Kurzweil says. Kurzweil idolized his father, and became obsessed with developing ways in which his father might be brought back to life. (He now believes it will be possible.) Two decades ago he began writing books, starting with The Age of Intelligent Machines in 1990, followed by The Age of Spiritual Machines in 1998 and The Singularity Is Near in 2005. He also has a new book out, Transcend: Nine Steps to Living Well Forever, which tells you how to stay healthy long enough to experience The Singularity and become immortal.

Kurzweil also likes to make predictions, and he claims he's found a foolproof, data-driven methodology for predicting the future. In 1990 he predicted that a computer would defeat a world chess champion by 1998. In fact, it happened in 1997, when IBM's supercomputer, Deep Blue, defeated Garry Kasparov. Kurzweil also predicted the rapid growth of the Internet and World Wide Web, and the ubiquity of wireless Internet access.

But hold on a minute. Who didn't think the Internet was going to catch on? And when you go back and check Kurzweil's previous books, you find that many of his predictions turned out to be wrong—not just a little bit wrong, but wildly, laughably wrong. During the height of the dotcom boom in 1998, Kurzweil predicted that the economy would keep on booming right through 2009 (and on to 2019, for that matter) and that one U.S. company (he didn't say which) would have a market capitalization of more than $1 trillion. Not even close. Kurzweil also predict-ed that by 2009 a top supercomputer would be capable of performing 20 quadrillion operations per second (20 petaflops in computer jargon), the same as the human brain. In fact, the top supercomputer just broke the one-petaflop mark—though Kurzweil says he considers all of Google to be a giant supercomputer and that it is, indeed, capable of performing 20 petaflops. Kurzweil also predicted that by now our cars would be able to drive themselves by communicating with intelligent sensors embedded in highways, and that speech recognition would be in widespread use. Neither has happened, but he insists they're both right around the corner. ("I was off by a few years," he says.)

Kurzweil makes predictions based on a notion that he calls "the law of accelerating returns," which holds that technology does not advance in a linear fashion but rather at an exponential rate. It's the difference between 1-2-3-4-5 and 1-2-4-8-16. Go out 10 steps and the linear string has reached 10, while the exponential string is hitting 512. With an exponential progression, at first, when the numbers are small, the progress doesn't look like much. But each new breakthrough enables the next breakthrough to occur more quickly, so the rate of change accelerates. Represented on a graph, the line of progress looks like a hockey stick—it's flat for some years, and then there's a sudden rise, which gets misinterpreted as a sudden breakthrough when really it's just the continuation of an exponential progression, Kurzweil says.

He cites as an example the work of the Human Genome Project. In 1990 scientists had managed to transcribe only one ten-thousandth of the genome over an entire year. Yet their goal was to sequence the entire genome in 15 years. After seven years, only 1 percent had been sequenced. But, in fact, the project was on track. The rate of progress was doubling every year, which meant that when researchers finished 1 percent they were only seven steps away from reaching 100 percent. Indeed, the project was completed in 2003. "People thought it would take centuries," Kurzweil says, because they foolishly believed that technology could advance only in a linear fashion. That same kind of linear thinking fuels the current hysteria about global warming. "People are assuming that nothing will change in the next few decades. They're ignoring the progression in renewable energy," Kurzweil says. After studying the subject, he and Google's Page concluded that the nanotechnologies needed to collect the energy of the sun are advancing at such a pace that in 20 years, solar power will be able to provide 100 percent of the earth's energy needs.

Apply that same kind of exponential progression to computer science, artificial intelligence, nanotechnology and biotechnology, and you arrive at The Singularity. Right now human brains are still much better than computers at doing things like pattern recognition. That's because while the brain works at slower speeds than a computer, it has 100 trillion interneuronal connections, so it can perform 20 petaflops, while the fastest supercomputer can perform one or two petaflops. But computers are doubling in power every year, and learning to do more things in parallel. Meanwhile, scientists are figuring out how the human brain works. Within two decades, Kurzweil believes, scientists will be able to "reverse engineer" the human brain and re-create its functionality in souped-up silicon. By 2029 a computer will achieve intelligence equivalent to that of a human being, or so close that the two cannot be told apart.