"Companies have got to learn to eat change for breakfast." Tom Peters
Ray Kurzweil is one of the world's leading inventors, thinkers, and futurists. Called "the ultimate thinking machine" by Forbes magazine, he was selected as one of the top entrepreneurs by Inc. magazine, which described him as the "rightful heir to Thomas Edison." He was the principal developer of the first omni-font OCR software, the first print-to-speech reading machine for the blind, the first CCD flat-bed scanner, the first music synthesizer capable of recreating orchestral instruments, and the first commercially marketed large-vocabulary speech recognition technology. He has founded and developed nine successful businesses in various areas of artificial intelligence, and his website, KurzweilAI.net, is a leading resource on A.I.
He is an inductee in the National Inventors Hall of Fame and recipient of the National Medal of Technology, the Lemelson-MIT Prize, 12 honorary doctorates, and awards from three U.S. presidents. He is the author of four previous books: Fantastic Voyage: Live Long Enough to Live Forever (coauthored with Terry Grossman, M.D.), The Age of Spiritual Machines, The 10% Solution for a Healthy Life, and The Age of Intelligent Machines. His new book is titled The Singularity Is Near: When Humans Transcend Biology.
tompeters.com asks ...
What is the singularity?
RK: The term singularity is a metaphor borrowed from physics, referring to an event horizon that's hard to see beyond. In physics, it's a physical event horizon surrounding a black hole. In this case, it's a future event in human history, which will be so profoundly transformative that it's difficult to see beyond. It refers to an amplification of our intelligence through our merger with the non-biological form we're creating.
There are a number of steps to understanding this. The first one is recreating human intelligence in a machine, with all of its subtlety and suppleness and pattern-recognition powers and emotional intelligence and so on. That involves capturing both the hardware and the software. Iâ€™ll talk about each of those, but first I need to address the overall acceleration of anything having to do with information.
Information technology in all of its forms is doubling in power—price-performance, capacity, bandwidth—every year. This is not just Moore's Law, and it's not just electronics. The amount of genetic data we're sequencing is doubling every year. The cost of sequencing a DNA base pair has come down by half every year—ten dollars in 1990 and a penny today. The genome project was controversial because we had only sequenced 1/10,000 of the genome in 1989 but it was done on time because it kept doubling every year. It took us 15 years to sequence HIV; we sequenced SARS in 31 days. The spatial resolution of brain scanning is doubling every year. The size of the internet is doubling every year. There are hundreds of examples of this.
So let's address first the hardware side of human intelligence.
When you say hardware, is that our brain?
RK: Right, that is the memory and processing capacity of the human brain. I have a number of different analyses from different people, including myself, that have shown that it will require between 10-to-the-14th and 10-to-the-16th calculations per second to functionally recreate the capabilities of the human brain. So to simulate realistically all regions of the human brain—several hundred of them—we'll require that capacity of computational ability. We will have that in a supercomputer by the end of this decade, and we'll have it in a thousand dollars worth of computation by 2020.
The next issue then is the software, without which we'll just have very fast calculators. We need to understand the principles of how human intelligence works. For that, we look at another grand project, which is reverse-engineering of the human brain itself. Thatâ€™s similar to where the genome project was ten years ago, which is to say it's expanding exponentially. We've made more progress than people realize.
We're just now gaining the tools to actually do it. For the first time, brain scanners are of sufficiently high resolution to actually see individual interneuronal connections signaling in real time. Until recently, technologies like fMRI could see only large groups of cells. They could tell somewhat where things were happening but not with enough detail to actually figure out the principles of operation.
Weâ€™re now getting the data, and weâ€™re also showing that we can turn this data into working models and simulations. For example, 15 regions of the auditory cortex have been modeled and simulated on a computer. The developers have applied sophisticated psychoacoustic tests to the simulation and get very similar results as applying the same tests to human auditory perception. There's a simulation of the cerebellum, which is where we do our skill formation, and the cerebellum comprises more than half of the neurons in our brain.
So the progress is already more than people realize, and it's exponential, not linear. A lot of observers, including otherwise sophisticated ones, misconstrue this. The pace of progress will continue to speed up just as it did for the genome project.
So itâ€™s that old issue that nothing happens as quickly as we think it's going to in the short term, but more happens in the long term than we can imagine.
RK: Yes. In the very short term, things may take a little bit longer than we expect, but people are projecting linearly far into the future. In the book, I talk about how radically an exponential trend separates itself from a linear trend. Most people think linearly. Itâ€™s human intuition to think progress will just continue at the current pace. But the pace of progress is accelerating and the growth of these tools in every area—computation, communication, brain understanding, our biology—is doubling every year. Thatâ€™s a factor of a thousand in ten years; it's a billion in 30 years. Well, actually 25 years because of a second level of exponential growth. Information technology is already profoundly affecting every aspect of our lives, and if you imagine it being multiplied by a billion over the next 25 years, that's rather transformative.
Coming back to human intelligence—we will have the software models, the principles of operation of all the regions of the brain, by the late 2020s. We'll then be able to apply it to computers, which by 2029 will be a thousand times more powerful than the human brain in a thousand dollars worth of computation.
You might ask what's the big deal with that, capturing human intelligence in a machine? We already have human intelligence—there are six billion of us—so why do we need more? We'll be able to combine the current advantages of human intelligence, principally our pattern-recognition capabilities, with ways in which machines are superior. For example, computers can share their knowledge at electronic speeds. It took us years to teach one set of computers to understand human speech. We actually taught the computer like a child, correcting its errors. After years it did a commercially acceptable job of recognizing speech. Now, if you want your personal computer to do the same thing—
Right. You just transfer the software.
RK: It takes us years to transfer knowledge to each child. But with nonbiological intelligence, you can just load the evolved patterns of one computer that has learned its lessons. So computers can share their knowledge at speeds that are a million times greater than human language speeds. And computers will be able to continue to increase, both the hardware and the software. The hardware will continue to grow by a factor of a thousand in less than a decade, and the software will also improve because these machines can go out on the web and master all of civilization's science and engineering. They can read their own source code and improve it in an iterative design cycle that will get faster and faster.
So the hardware and software will continue to evolve very dramatically, whereas human intelligence is effectively fixed.
But why is human intelligence fixed? Are we limited by the size of our brain?
RK: It's limited by the size, and more importantly it's limited by the paradigm of using signaling, using chemical gradients that move at a few hundred feet per second, which is a million times slower than electronics. Now that we can actually see individual interneuronal connections, we see that not only does our brain create our thoughts, but also our thoughts create our brain. When we think about something, we create new connections. That is part of the paradigm of how human intelligence works.
But there's a fixed architecture. There's a fixed size, a fixed set of regions that communicate, a fixed way of processing information. The brain doesn't just redesign itself using a communication method that's a million times faster. Of course, we ultimately will do exactly that, but through technology, not biology. We have, by my estimate, 10-to-the-26th power calculations per second in human civilization, in the six or seven billion humans. Fifty years from now the biological portion of our intelligence will still be 10-to-the-26th power. The non-biological portion, however, will be dramatically greater because it is growing by a factor of over a thousand every decade.
Another point is that this is not an alien invasion of intelligent machines. Computers are already a very important part of our civilization. If all the A.I. programs in the world were to stop tomorrow, our civilization would grind to a halt. You couldn't get money from your bank; transportation and communications would stop. There are hundreds of A.I. programs deeply embedded in our economic infrastructure, designing and producing products, controlling communications, and so on.
Ultimately the non-biological portion of our intelligence will greatly exceed the biological portion. I put the singularity at 2045, because at that point the non-biological portion of our intelligence will be a billion times greater than the entire biological portion. That's a profoundly transformative change. In my mind, that's still human intelligence. I don't like the word transhumanist, because it means going beyond human. This is still the human civilization in my view. It is transbiological. The book is about transcending biology, not transcending our humanity.
Your book is daunting at some level, but I think you make very strong cases for the facts. But what does the singularity look like on the planet? Is this just a select group of people?
RK: It's not a select group. If you look at technology today, only the wealthy can afford it when itâ€™s first introduced, and at that point it doesn't necessarily even work that well. A few years later, it works a little better and it's expensive, but affordable for more people. Eventually, it becomes inexpensive and works quite well.
So early adopters are VC personnel in a way.
RK: Yes. And it's very experimental at that point. Look how inexpensive cell phones are now. Look at the Asian societies. The World Bank reported that poverty in Asia has been cut in half and will be cut by 90 percent within ten years. In many of these societies, most people were pushing a plow 15 years ago, and now they're thriving information economies.
Right. But that doesn't seem to be the general perception.
RK: The general perception is wrong.
But clearly this is one of the hurdles you're encountering in your thesis, isn't it? Peopleâ€™s perceptions of what you're talking about.
RK: There is a sort of dystopian perception of machines out there, but at the same time, thereâ€™s a strong belief in progress and the ability of technology to solve problems. You see both things at the same time. That process of going from early adoption of unaffordable technologies to ubiquitous use of very inexpensive technologies takes ten years today, but ten years from now, it will take five years, and 20 years from now, it will take only two or three years. This refers to another trend of doubling the paradigm shift rate every decade.
So these technologies actually will move very quickly. People are often very myopic in seeing these kinds of trends. Look at AIDS drugs. That's an information technology. Fifteen years ago, it cost $30,000 dollars per patient per year for drugs that didn't work very well. Now it's down to $100 dollars per patient per year in poor countries for drugs that work quite well. So it's actually affordable now to provide these drugs for sub-Saharan Africa. Individual African patients can't afford it yet, but it's now feasible that organizations like the Gates Foundation and NGOs and governments can actually provide them.
So the faceless forces of the Law of Accelerating Returns move in the right direction. At any point in time, there's more that we can and should do to bridge the have/have-not gap. It's tragic we didn't do more about AIDS; millions of people lost their lives because of inaction. But with all the will in the world, there's only so much we could have done with drugs that cost $30,000 dollars per patient per year and don't work very well, given these very poor societies.
So the technology moves in the right direction.
This conversation brings to mind a scene in The Matrix, where Trinity has to fly a helicopter, but she doesn't know how. So in an instant, she downloads information that makes her able to fly. Is that part of this singularity? Is that sort of a semi-reality for 2050?
RK: Aside from its dystopian nature, The Matrix introduced a number of concepts that make sense. The basic idea of full immersion in virtual reality is quite realistic. You asked what the singularity will look like. To its participants, it will look pretty fantastic, because one of the applications of nanobots in the brain will be to provide full immersion virtual reality within the nervous system. If you want to go into a virtual environment, the nanobots shut down the signals coming from your real senses and replace them with the signals that you would be receiving in the virtual environment. To your brain, it feels like you're in that virtual environment, so you can be an actor in that environment.
The design of virtual reality environments will be a new art form. Some of them will be earthly recreations, such as a Mediterranean beach, and some of them will be fantasies. You can have a different body. You don't have to be the same person.
We have a very early version of that now with this Second Life game. A lot of people are living a virtual life in this place at secondlife.com, and people are paying real money for virtual property.
RK: Right now we're using a very crude audiovisual virtual reality that sits on the two-dimensional small screen. We'll have full immersion, three-dimensional visual-auditory virtual reality in the next decade even before we go inside the nervous system. You and I are now in an auditory virtual reality environment, but by the late 2020s we'll be able to do this non-invasively from within the nervous system. That's really more like a full Matrix concept.
So, like any other technology, these things won't be perfect when they're first introduced, but they ultimately will be very competitive and have a lot of advantages. Youâ€™ll be able to interact with anyone, regardless of where you are.
And you'll always have the delete button, won't you, if things start going drastically wrong?
RK: Yes, that's a big advantage. We can hang up on a phone call, which is auditory virtual reality. So, for example, virtual sex is safer. There are no STDs, no risk of pregnancy, and in case of interpersonal violence or other discomforts, you can always leave the experience. You'll have those safety nets built in.
Ultimately we'll be spending lots of our time in these virtual reality environments. Once we have full nanotechnology and can create physical things very quickly, we'll be able to bring some of those rapid morphing capabilities into real reality as well.
I've read that you're taking 250 supplements a day to rewire your body chemistry and that you've managed to eradicate type II diabetes from your biochemistry. Is that right?
RK: Well, this is another important issue, another aspect of what we're talking about, which is understanding human longevity. I had another book that came out last year, coauthored with. Terry Grossman, M.D., called Fantastic Voyage: Live Long Enough to Live Forever. There we talked about the impact of these technologies specifically on health and longevity. We talk about three bridges to radical life extension. You can get on the first bridge now, which is aggressively applying today's knowledge, far beyond the watered-down public health recommendations.
We baby boomers have to really be aggressive to be in good shape about 15 years from now when we have the second bridge, which is the full flowering of the biotechnology revolution. The whole new paradigm in biology now is to model biological processes, like arthrosclerosis, cancer, or insulin resistance, and reprogram those processes. We have means of turning genes off with RNA interference. There are new methods of adding genes that get the information in the right place in the right chromosome. We can turn enzymes on and off.
Just to give you one example, Pfizer's Torcetrapib turns off a key enzyme that's needed in a certain stage of arthrosclerosis. The phase II trial shows that it essentially stopped arthrosclerosis, and now they're spending a record $1 billion on the phase III trials. That's just one of thousands of these developments.
This full blossoming of the biotechnology revolution will come in about 15 years. It's also an exponential process. Then we will have very powerful tools to stop these disease and aging processes. And that's just bridge two.
That will bring us to the third bridge, which is nanotechnology, where we can go beyond the limitations of biology. The killer app of nanotechnology will be nanobots, blood cell-size devices that can go inside our body and brain and keep us healthy.
So our book talks about all three of these bridges, but more than half the book is on bridge one, including a practical guide to reprogramming your biochemistry today. We don't have the tools today that we'll have in 15 years, but if you put together the right program of nutrition supplements, you can, for example, dramatically lower cholesterol. My cholesterol was 280 but it's been 130 for many years. I had type II diabetes 22 years ago but itâ€™s been gone for 20 years. All my levels are close to ideal, but were far from that a couple of decades ago. And being an engineer, I don't just take these pills and hope for the best. I fly with an instrument panel. I take 40 or 50 measurements—blood levels and other levels on a regular basis.
I understand. It sounds extreme to me and, I imagine, to many others. Are there a lot of people on the planet doing what you're doing?
RK: There are quite a few people who are active in this kind of thing and we've had many very enthusiastic testimonials from people who have overcome major health issues using the information in our book.
It's not a one-size-fits-all program. The thesis is to understand your own situation. In my view, what's extreme is somebody having a heart attack. Or getting cancer. Or any of these major diseases. People may wax philosophically about how death is a good thing, but the reality is it's a terrible tragedy.
It seems in all of this discussion, there is this notion of living forever. Itâ€™s hard to wrap our current brains around that notion. We're accustomed to dying. We go to Woody Allen movies that are all about dying, and that's what we think about. So to be able to think about not dying just seems too extreme for our measly little human brains at this point, right?
RK: Right. In Chapter Seven, "Ich bin ein Singularitarian," I introduce the idea of having a philosophy that takes into account these significant changes. I'm not trying to introduce a new set of dogmas, but in my mind a singularitarian is someone who has thought about these things. It's not a 20-minute proposition. If I present it to people who have never thought of anything like this before, they often run into issues and dilemmas. It's taken me decades to think this through even this far, so it's not an easy concept to grasp.
I think it's actually very liberating. While we don't have these technologies at hand yet, I think we can have a very high degree of confidence that we will have them. That knowledge can affect our behavior today. We can apply today's knowledge, which is already quite extensive, to enable us to be in good shape when these developments do come around.
We rationalize death and illness as good things because we haven't had any alternative. A lot of religion, for example, is devoted to rationalizing death as a good thing. There are lots of movies and literature about the "good death," which I think is an oxymoron.
But in effect, we're dying from the moment we're born, right?
RK: Well, you can say that you're growing from the moment you're born. And I believe we'll be able to continue to do that. We don't need death to give life meaning. What's really meaningful about life is the expansion of human knowledge. We're the only species that has knowledge that we pass down from generation to generation, and our ability to understand and appreciate knowledge is going to continue to expand exponentially. I think that's very good news. By knowledge I don't just mean dry databases; I mean music, art, science, engineering, literature, appreciation of history, and so on.
I'm thinking about the business world where there's always talk about knowledge management. People and companies are concerned that the baby boomers are going to retire and all of this information will leave with them. Is there some cause here for celebration? Because if everyone continues to live long, then each individual collects more and more information that may be passed along in this computational way.
RK: We're capturing more and more of our knowledge non-biologically. We're gaining more and more powerful tools to access that knowledge. And people ask, "How are people going to get used to these changes?" Look how quickly people adapt to changes in the world. Certainly at the Tom Peters organization you've got a front-row seat on that. People didn't use the word "blog" until two or three years ago. Only three years ago people said you couldn't make money in internet advertising. Now there's a company with a hundred billion dollar cap in which 99 percent of its revenues come from internet advertising. The first reference to the World Wide Web in the New York Times was in 1993. So we adopt new approaches very quickly. I don't think there will be any hesitation to adopt cures for disease. If you put something on the market that will slow down aging, it's going to be gobbled up very quickly. Aging is not a good thing; it's a loss of your faculties.
Yes, obviously the whole baby boomer generation is working hard to defy aging, although right now it seems to mostly be in the realm of the outside physical aspect, plastic surgery and all. Maybe that's the beginning.
RK: We're going to find better ways of doing these things.
You have a line somewhere that most inventions fail because the timing is wrong. But do you have some sense of timing with your body? Do you wish you had born a little bit later? Is that something you think about? I mean, you've put 2045 as this date out there for yourself.
RK: 2045 is a point of very profound transformation. But with the maturing of the biotechnology revolution, we will be able to fix the problems of biology within 15 years. There are very exciting things in the pipeline right now, and each year we've seen more and more things happening.
Having thought about this and studied it quite a bit and come out with two books that Iâ€™ve written in the last year on this subject, I'm very confident that I'm ahead of the curve. I've taken extensive biological aging tests. There's some controversy about them, but I think they do give us reasonable information. They determine your biochemical levels and other measurements, such as tactile sensitivity, memory, and reaction time, and then compare your levels to population norms at different ages. When I was 40, I came out at about 38 on these biological aging tests. Now I'm 57 and I come out at about 40.
So you haven't aged in the last 17 years.
RK: Well I've aged two years, according to these tests. I'm hoping to make it back to 38 before I hit a chronological 60. I think I'm aging very slowly.
And is regular exercise a part of this regimen?
RK: Definitely. That's very important. We've put together a lot of different ideas. Where our program is unique is in the aggressive supplementation, really reprogramming your biochemistry. On the one hand it's good to be natural, and there are certainly a lot of unnatural things in our food that are not good for us.
On the other hand, just being natural is not good enough. When our genes evolved it was not in the interest of the species for people to live to be even as old as I am. Once you were done with childrearing, you were just using up the resources of the tribe, and there were very few calories to go around. Life expectancy was in the twenties 10,000 years ago. It was 37 in 1800. We have to go beyond biology to really bring our biochemistry to a more useful state.
So that's what I'm doing. I can't be 100 percent confident; after all, accidents can happen, so I'm also careful when I drive around town.
But what about the possibility of some huge discontinuity? Do you think about that?
RK: Well, I am worried about existential risks of the downsides of these technologies. Chapter Eight in the book is called "The Deeply Intertwined Promise Versus Power of GNR"—genetics, nanotechnology, and robotics. There's been a lot of discussion about the dangers of bioengineering and self-replicating nanotechnology. I talked about these things in The Age of Spiritual Machines in 1999, and Bill Joy's Wired cover story, "Why the Future Doesn't Need Us," was written in reaction to that book.
Bill and I coauthored an op-ed piece in the New York Times a couple of weeks ago criticizing the publishing of the 1918 flu genome.
Why were you against the publishing of that?
RK: No one would advocate publishing the precise design for an atomic bomb. In fact, that is illegal, and the couple of times it's happened, there's been a big uproar and it's been taken down. But this is actually more dangerous because it would be easier to recreate the 1918 flu from this genome, which is a recipe, than it would be to create an atomic bomb. You don't need rare materials like plutonium. And even if you had the plutonium to make an atomic bomb, you'd need very exotic industrial processes, whereas you can send genetic sequences to mail-order houses and get them constructed for you.
But isn't the upside that this is available to everyone who might be working on a possible cure for what is looming around the corner?
RK: Yes, definitely. And that's why the government published it. What we advocated was that this information should be provided only to people who need it.
I see. Rather than publishing it to the world, just publish it to the limited number of people you know who are working on a cure.
RK: Yes. Provide it with some security assurances. It should be confidential, a secret. Why provide it to the terrorists of the world by putting it on the web?
So I'm concerned about the downsides. I'm optimistic that we will prevail over these existential risks. I'm not as optimistic that we won't have painful episodes. I think we probably will.
Yes, it would be an unnecessarily utopian vision to think this would all go smoothly. Nothing has in the past, right?
RK: Right. But it's also the case that these major disruptions haven't disrupted technological trends. Fifty million people died in World War II, but the exponential growth of price performance computing sailed smoothly through the entire century, through up and down, thick and thin, depressions and booms, war and peace.
I'm somewhat convinced that we are a warring species. Is that going to change in the singularity, or will we just fight about something different in the future?
RK: I'm not predicting that we're going to get rid of human conflict. I think we'll have more insight into ourselves and be able to identify some sources of human dysfunction, but I'm not predicting that will enable us to overcome all human conflict.
I do believe that human conflict is decreasing. It may not appear that way since we now experience human conflict more acutely. In World War II 20,000 people could die in a battle, and maybe there would be a grainy newsreel shown at the movies two weeks later. Now we have a front-row seat for everything that happens. There have been analyses showing that war is actually decreasing. I think the reason for that is that all this decentralized electronic communication is very democratizing. When I wrote about that in The Age of Intelligent Machines, I predicted that the Soviet Union would fall apart because they wouldn't be able to control information. In fact, that coup against Gorbachev in '91 fell apart because of this clandestine network of fax machines and email over teletype machines. Grabbing the centralized TV and radio station didn't work anymore to keep everybody in the dark. And that communication network has grown enormously with the internet and with blogs. You've got millions of blogs, even in China.
I was at a pop tech conference recently where Rebecca McKinnon from Harvard noted that all of the Chinese search engines are coming with built-in censorship. For instance, if you're in China and you search on Tiananmen Square, you don't get any results.
RK: Yes. But it's a pretty narrow censorship.
She sees it in some way replicating itself because now American companies are developing software for the Chinese market, and they are hard coding the censorship in. So in a way it's becoming part of the base code.
RK: But it's not stopping the flood of communication.
I don't know if that's true.
RK: There's certain sensitive information, which people understand they can't talk about, but what is that? One percent of the things people want to talk about? It's not changing the flood of democratization, this enormous power of decentralized communication. They're not stopping the internet; they're just narrowly censoring certain issues.
You don't see it as a big issue?
RK: I certainly don't support it, but I don't see it as an issue that's really changing the nature of this trend.
That's fair enough.
RK: Another issue people bring up is stem cells. Isn't that a case of government opposition stopping the progress of biotechnology? I think it's very similar. These are stones in the river. Progress flows around them. Even in the narrow area of biotechnology called transdifferentiation, which is what the stem cell issue is all about, there's been tremendous progress. I support stem cell research and am opposed to the restrictions, but aside from the ethical and political issues, we don't ultimately want to use embryonic stem cells anyway because there are so few of them and the DNA doesn't match the recipient. I want to create pluripotent cells from my own skin cells, a process which has recently been demonstrated. The point is that opposition to stem cell research is not by any means halting progress in biotechnology.
Thank you, Ray. I appreciate your time today. I've always wanted to talk to you about one of your books.
RK: I enjoyed the dialogue. It was fun.
The Singularity Is Near: When Humans Transcend Biology
Fantastic Voyage: Live Long Enough to Live Forever (coauthored with Terry Grossman, M.D.)
The Age of Spiritual Machines: When Computers Exceed Human Intelligence
The 10% Solution for a Healthy Life: How to Reduce Fat in Your Diet and Eliminate Virtually All Risk of Heart Disease and Cancer
The Age of Intelligent Machines