Nerdworld.com | Nerd of the Week: Ray Kurzweil
November 6, 2000
Source: Nerdworld.com — November 6, 2000 | Morgan Michaels
RK: By the age of five I was convinced I would be a scientist/inventor. I always knew what I was going to be. I got involved in projects [that] grabbed my imagination and that usually winded up being a lot more complicated than originally anticipated. At the age of seven or eight, I built a puppet theater that had mechanical systems that changed the scenery. I got involved with computers at about the age of twelve. I built some computer like devices with electrical parts I would buy down on Canal Street in New York. You could buy these telephone relays that were primary electrical devices I could use to implement logical circuits.
NW: So you built a whole computer?
RK: I built this computer-like device when I was twelve for a junior high school science fair. I also started programming in Fortran. So when I was about 14 and 15 I built another computer — my first pattern recognition project. I programmed it to analyze melodies of a particular composer and then compose original melodies in that style using pattern recognition principles. That was my Westinghouse Science Talent Search Project (winners went to Washington and met President Johnson). After that [it went] to the International Science Fair, where it got first prize.
NW: At the age of fourteen, in the sixties, how’d you get access to computers to do Fortran programming?
RK: I had a job working for a research institute affiliated with one of the universities… I think it was New York University in New York. It was a summer job for a couple of summers. I started out doing these four-way analyses of variance on these old electromechanical calculators that actually had these calculating algorithms built mechanically. Someone could actually do square roots … put the number in and it would actually grind away very noisily for like three or four minutes. But I quickly got tired of it…. Four-way analysis of variance involve all these repetitive calculations. We filled data in by hand in these various tables and followed this elaborate algorithm using a calculator and using pen and paper spreadsheets.
So they actually had access to an IBM 1620 — forerunner of many computers — which wasn’t a large mainframe. I could use it particularly in the evenings I had sort of free access to sit on it and write up the program in punch cards and key in the Fortran and [the computer] would compile it. Then I actually programmed it for four-way analysis of variance with unequal subset cell frequencies. IBM ended up distributing it to the 1620 and it was used for a while by similar research institutes.
NW: Was there anyone along the way who helped inspire your Nerdliness?
RK: Well, there was Marvin Minsky, with whom I corresponded when I was in high school. He invited me to come up here and visit him at MIT when I was in high school. He was and remains a very generous teacher in terms of his time. I think the thing he enjoys most is spending time with students. He spent a lot of time and was really interested in things I was doing. He wasn’t even at that time a legendary professor in AI [and it] was not a very well known field. It wasn’t a field I was interested in (this was the early sixties).
I also visited Professor Rosenblatt at Cornell who was doing work with perceptrons, an early form of neural net. This was research that ultimately was negatively affected by the publication of Minsky and [Seymour] Papert’s book Perceptrons in 1969… Minsky had done early work in connectionism and connectionist architectures. He was, in the early 60′s, part of what then was traditional AI school of thought. Both of those professors turned me on in that they showed an interest in what I was doing and really seemed not to care that I was just a high school student.
NW: At any point did you consider getting into the music field?
RK: I was in a musical family and there was a strong musical influence, but I never considered being a musician. I did consider being a writer and in fact had a dual major in computer science and creative writing (at MIT). Lillian Hellman was a professor there and I studied with her.
When I went to MIT in 1965 they only had eight or nine computer science courses, which I took for the first year or year and a half. So then I started majoring in literature. My parents were unhappy about that–my parents were both struggling artists. Even at that time science seemed to be the wave of the future and considering a more artistic career was upsetting to them. But I kept up an interest in both. Through writing nonfiction I was able to invent with material and technological resources that don’t exist today; they exist in the future.
NW: I understand that you have an ongoing collaboration with Stevie Wonder?
RK: Yes, well I’ve known him for twenty-five years.
NW: Are you currently collaborating with other celebrities?
RK: I’ve had a lot of interaction with quite a few musical celebrities in the course of my work with Kurzweil Music, but I wouldn’t call them collaborations. There’s a musical group called Our Lady Peace that has shown an interest in my work and is going to be showing the influence of my book [The Age of Spiritual Machines] in their next album. But I haven’t had the level of collaboration that I’ve had with Stevie Wonder with any other musical stars.
NW: Are you still developing tools for musicians?
RK: I have a company called Kurzweil CyberArt Technologies and the purpose is to create software tools to help in the creation of art in the visual, literary and musical arts. The marketing model is viral marketing over the Internet of free versions of the software and then inexpensive upgrades over the Internet to premium versions for prices like $30. We have one product up there now that is Ray Kurzweil’s Cybernetic Poet that can write original poetry. But its primary purpose is to help you write poetry. You write poetry in one window and then it can give you lots of ideas based on its database of poetry from different poets. It gives you lots of suggestions [for writing original poetry] it’s designed to stimulate your imagination.
We’re going to very shortly be introducing Harold Cohen’s Aaron which is a cybernetic artist and that doesn’t so much help you create art, [it's] a cybernetic artist with its own personality. It’s something that Harold Cohen has developed over thirty years, it’s been his life’s work — he’s a brilliant computer scientist and artist. The output of Aaron actually hangs in museums around the world. There’s going to be a version that’s a free screen saver that paints paintings on your screen line by line. Every painting is different; you’ll never see the same painting twice. We’ll be launching that in a few months.
We [also] plan to have musical technology that would enable someone with music appreciation ability but not necessarily music playing skills to create their own music by jamming with their computer and will include a software-based synthesizer.
NW: One of the things about art that’s hard to duplicate is the shared experience with the artist. What’s your feeling about our experience of art and technology in the future?
RK: Right now… our most complex machines are still on the order of a million times simpler than the human brain. Although we’re making very strong progress in reverse engineering of the human brain — in fact we’ve made some very strong progress even since my book came out two years ago–we still don’t have maps or reverse engineering of 99% of the human brain. We’ll get there because brain scanning, our ability to reverse engineer the human brain, as well as computation, communication and miniaturization are all growing exponentially.
These predictions are not just pulled out of the air. I have a detailed model that I’ve been developing for a couple of decades on how these particular capabilities evolve. The model has actually proved quite predictive through the nineties and [I] believe it can anticipate what kind of capabilities will be possible.
One of the key features of the future that most observers, including some very thoughtful ones, don’t appreciate is that the rate of progress itself is accelerating. I had a debate a few weeks ago about Bill Joy’s article in Wired where this Nobel Prize winning biologist said that there’s no way we’re going to see self-replicating nanotechnology for at least a hundred years. I pointed out that’s actually a very good estimate of the amount of technical progress required to achieve that technical capability at today’s rate of progress.
But the rate of progress is not a constant. According to my model, it’s doubling every decade. So we’ll see one hundred years of progress at today’s rate of progress in 25 calendar years. Most people don’t really appreciate the sort of revolutionary implications in the fact that progress itself is accelerating. So we will see entities of the complexity and depth of the human brain in about 30 years. We already make use of information from the reverse engineering of the human brain. [Eventually,] machines will either be copies of the human brain, suitably modified or extended, or new systems based on our understanding of how the brain works with computer hardware that’s millions, ultimately billions times more powerful than what we have today.
So the attributes that we now associate with human intelligence — the ability to respond appropriately to emotion, to communicate in art, which is the ability to communicate philosophical and emotional ideas from an artist to an audience — will be feasible for non-biological entities. One of the key predictions of my books is that ultimately there won’t be a clear distinction between biological and non-biological intelligence. We’re going to be expanding our own intelligence by amplifying it, [and] merging it through intimate connection with our technology.
I describe a number of scenarios where that can be done non-invasively, for example sending intelligent nanobots to the brain through the capillaries. And you can have billons of nanobots all merging their own intelligence by being on a distributed local area network, communicating wirelessly with each other, and also communicating directly with our neurons through wireless communication. That capability exists already in a crude form. Attributes we now associate with human intelligence will be ones we associate with non-biological intelligence. And there won’t be a clear distinction. When you talk to a human of biological origin some portion of their thinking — deeply intertwined and merged with their biological thinking — will be non-biological.
One aspect though of non-biological intelligence is that it’s growing exponentially. We have 10 26 power approximately of calculations per second of biological thinking among human beings on earth today. And that’s not going to change appreciably. Our non-biological thinking is at least 6 or more quarters of magnitude less than that today but growing exponentially and will surpass biological thinking before 2030, and then continue to grow exponentially. By the middle of the century the bulk of thinking on the planet will be non-biological but nonetheless derivative of human intelligence. It’s really an expansion of the human civilization; kind of the next step in evolution, which is proceeding not through DNA-guided evolution but through our technology-guided evolution.
NW: Getting off the subject for a bit, what do you feel is the future of large-scale projects?
RK: There’s room for all size projects [and] there’s a tremendous amount of capital available for all size projects. Certainly there’s a tremendous explosion of little companies that are exploring all kinds of creative ideas. As some of them show viability they have opportunity to attract very large sums of capital. The Internet together with emerging computational resources are definitely going to transform all the old business models. Virtually every company has had to restructure itself, so the entire economy is at the disposal of these new technologies. There’s plenty of opportunity to do very large-scale projects. But there’s also the ferment of tremendous amounts of creative effort by thousands of little projects. The real scale of it is thousands of times greater than it was twenty-five years ago when I first started.
NW: What low-tech devices do you still use that people might not expect you to use (for instance, a Filofax instead of a PalmPilot)?
RK: I keep my to-do list on a piece of paper. I word process it periodically. I have it on several sheets of paper. I highlight them, I check them off, and I write some new items in pen. Maybe once a month I’ll go back and update this file… I tried a PalmPilot, which I though was hopelessly slow at data entry. I think when we get viable voice recognition on these hand-held devices it might be viable but it’s just incredibly slow at data entry. The device is just too slow and the screen too hard to read. I do make a habit of trying new technology just as part of being a student of technology. I just didn’t find the PalmPilot satisfactory.
NW: What do you feel are your responsibilities are to Nerds Worldwide?
RK: I try to do projects that have worthwhile cultural or social purpose. What excites me is not so much abstract formulas. What excites me is the sort of link between dry formulas on a blackboard and actually having some transforming effect on people’s lives. The projects I’ve taken on are focused on changing some aspect of our social or cultural lives. I’ve also tried to build organizations where people really enjoy working… We try to put together teams where everybody loves to be a part of it and are devoted to the mission and who work well together. Teams we’ve put together are still together.
NW: Do you have any special clothing, music or other talismans that you use to inspire your creative work?
RK: Well, I’m not usually wearing anything when I do my creative work [laugh]. I [do] have a specific approach to doing creative work that I’ve used for several decades. When I go to sleep I assign myself a problem, it can be any kind of problem. It could be some algorithmic solution to some formulas, it could be some business strategy problem, it could be an organizational problem, and it could be an interpersonal problem. I specifically try not to solve [the problem] because that would drown the creative process, but I do try to think, “What do I know about this? What attributes would a solution have? What am I looking for? What are some of the constraints?” I just review what I know about the problem and then I go to sleep.
Two things happen. One, when I get up I’ll invariably have some new insight into the issue. Freud said that when we’re dreaming, our senses are relaxed and we basically shut off those portions of our mind that tell us what we’re not allowed to think about. So a lot of taboos will emerge in our dream and we’ll do and think things we don’t allow ourselves to do while we’re awake; sometimes we don’t even remember them because the taboos are so strong against even thinking about some of these things. Some of these taboos [impose] constraints on why you can’t solve a problem a certain way. A lot of professionals in every field — engineers, doctors, lawyers — learn to shut down certain creative approaches to problems due to the constraints of professional thinking in their field.
By thinking about problems without those senses you can really find creative solutions to them. But what’s really effective is that period of time when you’re awakening. You realize you’re conscious and you could will yourself to get up, but you choose not to and you’re really still in a dreamlike state. It’s a period of lucid dreaming. You dream can even continue, but you now also have your conscious faculties.
One of the advantages of this dreamlike state is that your senses are relaxed, but the disadvantage is that your rational faculties are not operating. In this lucid dreaming state you have both. You’re conscious enough to have your logical faculties but you’re also still in the dreamlike state where you’ve relaxed the restrictions to creative thinking.
I really try to stay in that state as long as I can. I find it [to be] a fantastically creative time. I’ve come up with inventions in that state and solved difficult problems, whether they were formulas or business strategies, I’ve written speeches in a few minutes in that state. I can really get a tremendous amount done and have great clarity of thought. So then after I do get up, all I’m doing is really just carrying out the decisions I’ve made in this dream state. I would say most of my creative thinking is done in that state.
NW: And now for a taste of James Lipton…Favorite word?
NW: Most hated word?