Technology Review (Jan/Feb 2000)
The Story of the 21st Century
Ray Kurzweil created a hubbub with his idea that we will soon be able to “download” ourselves into machines and live forever. Find out what else he’s got up his futuristic sleeve.
Struggling to find time in a busy schedule for yet another interview, Ray Kurzweil jokes that the media frenzy surrounding him these days only happens “every thousand years.” That’s because the inventor and entrepreneur who brought us such products as the Kurzweil electronic keyboard, a text-to-speech reading machine for the blind and voice-recognition software is also one of the most audacious—and, some say, accurate—futurists around. Kurzweil’s fearlessly detailed predictions make his latest book, The Age of Spiritual Machines, a must-read for the turn of the millennium.
TR Associate Editor Rebecca Zacks visited Kurzweil at the Wellesley Hills, Mass., offices of Kurzweil Technologies, one of the half-dozen high-tech companies he has founded since selling his first major enterprise to Xerox in 1980. The son of a composer and trained as an MIT undergraduate in both computer science and creative writing, Kurzweil moves deftly from music and art to computer processors and nanotechnology to immortality, evolution and God. And though his manner and his gray pinstriped suit are remarkably subdued, his ideas about the future are explosive.
TR: You’re accomplished both as an inventor
and as a writer—how do you see those two roles fitting together?
KURZWEIL: In writing, you’re also inventing. My main interest is to write about the future, though I did write a health book. Lately, my interest in health has intersected with my interest in computers, because they both have a bearing on the issues of longevity and immortality—keeping our biological bodies and brains healthy is the first bridge to immortality. That’ll bring us to the bioengineering revolution. Within 10 years, bioengineering will extend human life spans at least a year every year. And that’ll be the second bridge that’ll bring us to the nanotechnology artificial intelligence revolution, which gives us a real shot at immortality. But writing about the future and technology is also an invention process, because you have to invent the future to have a compelling statement about it.
TR: One of your more dramatic statements is
that in the second half of the 21st century we’ll routinely be able to scan a
person’s brain and “reinstantiate” that person in a computer—no more squishy
human body necessary. What did you have to mentally invent to come up with that
KURZWEIL: People ask, “How is that possible, scanners really can’t resolve to that resolution,” so I came up with the idea of scanners that would scan the brain from inside. We already have scanners that can scan with extremely high resolution, providing the scanning tip is right next to the neural features. Well, how are you going to move the scanning tip to every point in the brain without destroying the brain? The answer is to send them in as nanobots, blood-cell-sized robots with little scanners that would travel through every capillary in the brain. How would they communicate with each other? They would all be on a wireless local area network, and they would use distributed processing, thus the computational ability of the nanobots would be assembled into one distributed parallel computer.
TR: Is this something we could do today?
KURZWEIL: Every aspect of it is feasible today, except for the size and the cost. And that’s where what I call the “Law of Accelerating Returns” comes in: There are all these accelerating technological processes that are increasing exponentially the cost effectiveness of computing. So the requisite computing for this scenario will be quite feasible within 25 to 30 years. What about size? Well, miniaturization is another exponential trend in technology, and right now we’re shrinking the size of technology by a factor of 5.6 per linear dimension per decade, so again, you can predict that this will be feasible within 30 years.
TR: What mistakes do technology forecasters
KURZWEIL: A lot of people when they talk about the future do it without any foundation or methodology. They, for one thing, are very afraid to look beyond one or two iterations of technology—for example that screens will be higher resolution, a little smaller, but then progress will stop. People very often fail to see the interaction between many different trends. If you look in different fields—brain scanning, nanotechnology, artificial intelligence, computing substrates, communication—you see how all of these fields are going to interact with each other to provide tools that in turn accelerate progress in each other. The most important element that’s very commonly missed, even by sophisticated futurists, is the accelerating pace of change. A futurist may make a statement that we won’t see a particular development for 30 years, because a comparable change took 30 years in the past. But time and the pace of progress is accelerating, so something that took 30 years in the past will only take 7 years in the future.
TR: How do you quantify the future pace of
technological progress so precisely?
KURZWEIL: Over the past 20 years, I’ve developed a mathematical model for what will be feasible in terms of computational power and memory and miniaturization, so that I have some methodology for making these predictions. The predictions that I made in the ’80s about the ’90s were quite accurate. I predicted, for example, that the computer would take the world chess championship in 1998—I was off by a year.
TR: Is the accelerating development of
technology, in particular computer technology, inevitable?
KURZWEIL: The creation of more and more intelligent machines is an economic imperative. People sometimes say maybe we’ll get to a point where we’ll decide not to build these machines because they’ll be too threatening, but it’s not a realistic scenario—we’d have to repeal economic competition, free enterprise, capitalism, to stop that progression. Anytime anyone creates a machine that’s a little bit more intelligent, it takes over the market. There are tens of thousands of projects with the force of economic competition driving the whole process forward. It’s not a centralized decision.
TR: How will we get from faster and smaller
computers to computers with human intelligence?
KURZWEIL: Basic computational power is a necessary but not sufficient condition. There are a number of different scenarios for how we can organize the software of intelligence, but the most compelling one is that we have an example of an intelligent entity in our midst: It’s the human brain and it’s not hidden from us. We’re already down the path of learning about it. For example, we use the kinds of transformations the human brain does on auditory information in our speech recognition system. There’s going to be a tremendous incentive to learn about the brain, to learn the secrets of intelligence and then replicate those methods. If you ignore this resource of the human brain, then you might say that we’ll never figure out the software of intelligence.
TR: Part of your mathematical methodology
dictates that evolutionary processes expand exponentially, and that the
development of technology is itself an evolutionary process—can you explain?
KURZWEIL: Think about the evolution of life on earth. It took billions of years for the first cells to form, and then in the Cambrian explosion, paradigm shifts only took a few tens of millions of years. Then later on, we went from primates to humanoids in only millions of years, and then Homo sapiens emerged in only hundreds of thousands of years. And then it became too fast for DNA-guided protein synthesis to keep up the pace of progress, so the whole cutting edge of evolution on earth has moved to technology created by the technology-creating species. So in my view, technology is actually a continuation of the evolutionary process that gave rise to the technology-creating species in the first place.
We can describe evolution as a sort of essential spiritual quest. As we evolve, as matter and energy evolves, it creates entities that are more intelligent, more creative, more beautiful, more loving. These are all the qualities that we associate with God. God has been called infinite—infinite knowledge, infinite intelligence, infinite creativity. Evolution never really becomes infinite. It remains finite, but it does become very large, so it’s moving in that kind of spiritual direction.
TR: We’re asymptotically approaching God?
KURZWEIL: But never reaching it. By the end of the 21st century, nonbiological intelligence will be trillions of trillions of times more powerful than human intelligence. That’s hard for us to imagine, and maybe from the human perspective, it’s virtually infinite. But from a literal mathematical perspective, it’s still finite, so we can consider God as an ideal that evolution never reaches.
TR: It sounds like such an optimistic
vision. Are there no drawbacks?
KURZWEIL: I think there’s always a struggle between the constructive and the destructive forces of technology. Biotechnology today is a very good example: On the one hand, we’re at the very early stages of a burgeoning revolution that’s going to reverse disease and aging processes over the next five to 15 years. But there’s an obvious downside: The means and skills exist in a routine bioengineering laboratory to create a pathogen that would be more destructive than an atomic weapon. The technology we’re creating for the 21st century will be even more powerful. Because it’s self-replicating, nanotechnology will ultimately be able to provide anything in the physical world that we want, so if properly applied, it can meet all of our needs and desires and create fantastic wealth. But there are also enormous dangers to nanotechnology. Self-replication run amok would be a nonbiological cancer that would be even more destructive than a biological cancer.
I tend to come out on the optimistic side of the field, that overall technology creates a better world despite our sometimes feeling a romantic desire for the good old days. Richard Dawkins calls evolution “the blind watchmaker”—I think he should have called it the “mindless watchmaker,” because he was using “blind” to mean “mindless,” which is insulting to blind people. There’s no intelligence behind the process but yet nonetheless it created all the wonders of the natural world and created human beings. This next stage of evolution, which is technology, is a mindful watchmaker. So we do have actually the ability to guide that process, and therefore the responsibility to guide it in constructive directions.
TR: One area where your companies and
inventions have already played a role in guiding technological evolution is in
the arts, and your most recent piece of software, “Ray Kurzweil’s Cybernetic
Poet,” is a foray into electronic writing. How does it work?
KURZWEIL: It’s a system that reads poems from a particular author, and it creates a language model that describes how those poets create poetry, and then it can write original poetry in that style. Probably the majority of those poems don’t work fully. However it comes out with really terrific turns of phrase and very interesting lines of poetry, so we’ve packaged it as a poet’s assistant—you write a poem in one window and the Poet’s Assistant will give you ideas for alliterations, rhymes, half rhymes, the next word of your poem, and turns of phrase that are relevant to what you’ve written, and you can fill up the screen with these different suggestions. It doesn’t have human intelligence but it can do clever things with language, and help you to write a poem of even prose. It’s available as a free download at www.KurzweilCyberArt.com.
TR: What will happen as computers become
more artistically adept, and something like a cybernetic poet does reliably
write poems that work?
KURZWEIL: In order for a computer to create completely satisfactory art, it needs to have a human level of intelligence. And when a computer does have a human level of intelligence, it brings up the issue of who is human—some of these cybernetic poets and artists will think that they are. And once a machine achieves a human level of intelligence, it will necessarily soar past it, because machines already have certain advantages over human intelligence. One is the ability to share their knowledge by quickly downloading software. They inherently will be much faster than humans because their electronic circuits are already 10 million times faster and their memories are much more accurate. If you take those inherent advantages of machine intelligence and combine them with what are today advantages of human intelligence—our pattern-recognition capabilities and the tremendous breadth and subtlety of our intelligence—that’s a very formidable combination.
TR: Do you look forward to the future that
KURZWEIL: I do. I hope to be around working on it. I’m still on the first bridge to immortality, which is trying to take care of my biological body and brain in the old-fashioned way. I do think there are dangers, and the story of the 21st century hasn’t been written. We have to create the next stage in evolution and infuse it with human values—not that we have a consensus of what those are.
TR: Will you be one of the first to make
the jump and say good-bye to your squishy body?
KURZWEIL: It’s a difficult question. You could scan my brain while I’m sleeping and recreate this new nonbiological Ray Kurzweil, which could come to me in the morning and say, “Hey Ray, good news. We’ve successfully copied and reinstantiated your brain and body, we don’t need your old brain and body anymore.” I might see some flaw in that philosophical perspective. I’ll wish the new Ray well and I’ll probably end up being jealous of him because he’ll be able to succeed in endeavors I could only dream of, but I’m still here in my old biological body and brain. It’s not clear how one gets over that divide to the other side—I haven’t quite figured that out yet, but I do hope to see the era.