Inventing the Future

June 24, 2009

Welcome to T2A Chat as we meet one of the world’s leading inventors, Ray Kurzweil. He was principal developer of the first CCD flat-bed scanner, the first omni-font optical character recognition, the first print-to-speech reading machine for the blind, the first text-to-speech synthesizer, the first music synthesizer capable of recreating the grand piano and other orchestral instruments, and the first commercially marketed large-vocabulary speech recognition. Ray joins us from Boston, Massachusetts.

From Madan Mohan, India (email): I would like to know about your latest focus on the inner workings of the human brain and artificial intelligence. I think it requires a lot of knowledge about human activities.

Ray Kurzweil: Well, that’s a very good question and artificial intelligence progress so far, up until perhaps a few years ago has not benefitted very much from understanding the human brain because we really didn’t have the tools to see inside it. You probably see these colorful MRI pictures where one part of the brain lights up when it’s solving a particular type of problem. And that tells us where things are going on but those have not been high enough resolution to actually figure out what the messages are and that it’s being used. But the spatial resolution of brain scanning is doubling every year. The amount of data we’re getting is doubling every year. We’re also able to analyze specific neurons and develop mathematical computer models of those.

And there has been tremendous progress in understanding the human brain in recent years. There’re now twenty different regions of the brain that had been modeled and simulated. And simulation tested against human performance. These include the auditory cortex where we understand auditory information; visual cortex; the cerebellum which comprises more than half the neurons in the brain where we do our skills formation like pitching a flying ball; and fairly recently slices of the cerebral context which is where we do our rational thinking, our recursive thinking responsible, for example, understanding human language.One of my key themes is that information technology doubles in power every year.

This has certainly been true for computers but also true for projects like these: understanding the human brain. I make the case in The Singularity Is Near that within 20 years, by the late 2020s, we’ll have models and simulations of all the regions of the brain. That will give us, first of all, more insights into ourselves. And that has been the quintessential goal of the arts and sciences in the millennia but also give us the logarithms to create machines comparable to human intelligence. We already have a lot of artificial intelligence in use in our civilization.

That’s a fairly recent phenomenon. Every time you sent an email, intelligent logarithms root the information, intelligent logarithms help design products, built them in robotic factories, keep track of just-in-time inventories level. You get an electric cardiogram, it comes back with an automatic diagnosis from your doctor. I can list a hundred other applications. Even narrow AI is doing things that use human intelligence at human levels and beyond but for very specific tests. The narrowness is gradually going to get broader and broader as we get more insights into the human brain.By the late 2020s, I believe we will have computers that can operate fully at the human level and combine that with advantages that computers already have at being able to master billions of items of knowledge and accurately shared information at electronic speeds.


Lim, Hio Tiao, Philippines (email): Are we not going too far? From a religious viewpoint, it seems that Man has gone too far.We went to the moon even without totally discovering our very own planet.

Ray Kurzweil: Well, I would say that all the major religions in the world support the idea of progress at advancing scientific knowledge to solve human problems, for example, overcoming human diseases. The Catholic Church supports this as do all the major religions. We really have a belief in human progress to overcome human suffering deeply ingrained in our consensus on philosophy around the world. One of my points is that progress accelerates. For half a century for the telephone to be adopted, we adopted the cell phone in only seven years. I’m talking about adoption by a quarter of the American population as a guide. So the pace of change is getting faster and faster.

Ten years ago, most people didn’t even use search engines. Think about a world without search engines. That sounds like ancient history less than a decade ago. Think about a world without tweets, that was one year ago. So the pace of change is getting faster and faster, that’s the key theme. And one application of that is to overcome human suffering. Human life expectancy was 37 in 1800s. Shubert and Mozart died in their thirties because there was no sanitation and no antibiotics. Think of the music they could have created. It also expands our creativity.

The tools of creativity are now in everybody’s hands. If you just want to create a motion picture ten years ago, you have to be in a big Hollywood studio or motion picture studio somewhere in the world. Today a kid in their dorm room can create a full-length high-definition movie with their PC and a five hundred dollar camera. A couple of kids at Stanford, with a thousand dollar laptop and a little software as a dorm project, revolutionized search and created a company worth a hundred billion dollars. So the tool of being creative, which I think is a quintessential human activity, is supported by our religious philosophies, is in everybody’s hands.


Kohei Maenobo, Oxford, United Kingdom (email): How do you convince people who are skeptical about your predictions of the future?

Ray Kurzweil: Well, I think the reason for skepticism is that people think in a linear manner. That’s our intuition about anticipating the future. That’s actually hard-wired into our brains. But the reality of the progression of information technology is not linear but exponential. And that has been very reliable and continuous, going back, for example to the 1890s census. I put every computer for example in the last twenty years on a logarithmic graph and it’s a very smooth curve showing a progression of trillion folds increase in capability over that period of time.

When I was an undergraduate in MIT, we all shared one computer in a building. The computer in your cell phone today that sits in your pocket is a million times cheaper, a million times smaller, and a thousand times more powerful. That’s a billion fold increases in capability that we’ve seen as an undergraduate. And we will do it all again in 25 years. I’m not just looking backwards and fitting past data but making forward looking predictions for several decades. In the early 80s, I have predicted the emergence of a world wide web with communications tied in together, hundreds of millions of people to each other and to the best knowledge sources emerging in the mid to late 1990s. People thought that was ridiculous when the entire American defense budget can only tie together one or two thousand scientists. To talk about connecting hundreds and millions of people, people say maybe that will happen hundred years from now. But it happened right on schedule. And I make many predictions like that.

If you compare linear progress, which is our intuition, but it’s wrong, to exponential progress, which is the historical accurate trajectory of information technology, it makes a big difference. Take thirty steps linearly – 1,2,3,4,5..- I get to 30. But if I take thirty steps exponentially – 2,4,8,16…- I get to a billion. It makes a huge difference and yet our intuition is governed by a linear perspective. Even sophisticated scientists make that mistake. When the genome project was announced in 1990, mainstream scientists said there was no way you can do it in 15 years. We had our PhD students and our most advanced equipment around the world, we only collected one ten-thousandth of the genome in 1989. This is going to take centuries.

Half way through the project, seven and a half years later, the skeptics are still going strong, saying, “I told you this wasn’t going to work. Here half-way through the project, we only finished one per cent of the project. But then actually, it was right on schedule. Because you started out doubling little numbers, once you get to one per cent, you’re only seven-doublings away from a 100 per cent. And the amount of genetic data continues to double every year. Seven years later, a year ahead of schedule, the genome project was completed as I predicted. So the reason that my prediction seems startling to many people who haven’t really analyzed the data is because of this explosive nature of exponential growth.


Erin: So when did you begin to focus on forecasting the future and has that become more of a focus than inventing or are the two linked?

Ray Kurzweil: The two are definitely linked. In fact, I decided I was going to be an inventor when I was five when other kids wondered what they are going to be when they grew up. I always have this conceit that I know what I’m going to be. I quickly realized about 30 years ago that the key to being successful as an inventor was timing. Almost every inventor – 90 per cent of them – gets their project to work just as they envision them but still 95 per cent of those projects at least failed in the marketplace because the timing was wrong and the market was not ready. Or that they were too late, they missed the window of opportunity. Realizing that, I became an ardent student of technology trends.

Being an engineer, I gathered a lot of data, tried to underline the key underlying information properties of technology like the number of bits being moved around or that price of performance of computing and so on. I discovered something that actually was quite surprising to me, that these trajectories, if you measure the underlying information properties, like the number of nodes in the Internet or the number of bits of data we’re getting from brain scanning, many different measures like that, are following amazingly predictable trajectories. And those trajectories are, as I’ve mentioned, exponential. Doubling, with a doubling time, depending on what you’re measuring, the doubling time might be one year or 11 months or 14 months. Human knowledge for example is doubling every 14 months. With that, you can make predictions. I’ve been making accurate predictions like that for the last 30 years. And I used some to time my own technology projects and it’s still a primary application but a side effect is that I can make prediction about ten years from now, twenty years from now, describe what the world would be like. And that’s the basis of my book, speeches and so on.


Erin: You talked about knowing at the age of five when you were already recognizing, as you put it, your conceit, that you were smart to know at that moment, but did you envision at that time what you would invent?

Ray Kurzweil: Well, I remember at the age of five having this feeling that if I put things together in just the right way. I had a lot of different construction toys, not electronics yet. But if I put things together in the right way, you can create transcendent effects. I did not have that vocabulary at age five but I have this idea that you can actually solve major problems that people had. Pretty quickly, I have this idea that I could help people like the blind, the deaf and maybe when I was seven or eight, I didn’t figure out the idea on how to do that or what it would look like. But I had this feeling that you can recreate the world simulated and overcome its limitations with inventions.


Susan: What about this reading machine for the blind? Why make one? Who was in your life that inspired this or was there someone in your life that inspired it?

Ray Kurzweil: Well, the key technical area that I have expertise in is an area called pattern recognition, which is teaching computers to recognize patterns. It turns out actually that the heart of human intelligence is our ability to recognize patterns. That is actually our entire foundation of our intelligence. So I developed a number of pattern recognition programs when I was in high school. I created a computer that could recognize patterns and music and then write original music with those patterns.

In the early 70s, I created a software system that could recognize printed letters and any typed style. The state-of-the-art before that was the OCR – Optical character recognition – could only recognized one-typed style that had to be perfectly printed. So I developed one that could deal with degraded print, smudged print, any typed style and so on. It was kind of a solution in search of a problem. I didn’t know what this might be good for. So we did some market research and look at different applications.

But the galvanizing event is that I happened to sit next to a blind guy in a plane and he was telling me how blindness is not really a handicap. He represents his company, makes deals all over the world, he was doing that now.But then he corrected himself and said, “well, there is one problem, one handicap associated with being blind, which is the inability to access ordinary printed material.” He had to get someone to read it, there’s a problem with privacy and he always has to get someone to do that. So that’s really galvanized the decision to apply this technology to building a reading machine for the blind. Other than that, I really did not have a personal connection to blindness. That required two other inventions- there was no flatbed scanner in the world at that time.

So we created the first flatbed scanner. We created the first text-to-speech synthesizer, so that the recognized print could be spoken out loud. In 1976, we introduced the first print-to-speech reading machine for the blind. I stayed involved in the field since then. That (the 1976 reading machine) was the size of the washing machine, cost 50,000 dollars. We now have a reading machine – I’m now holding it in my hand – that fits in a 4-ounce cell phone. A blind person just snaps a picture of a page – it could be a sign on the wall, ATM display, a manual handed out in a meeting, and it would instantly recognize the page and read it out loud through the cell phone.


Erin: You mentioned a solution in search of a problem, so these things, the keyboard, the synthesizer, the speech recognition systems, did they really come about through problem solving as such or did some of them just come to you?

Ray Kurzweil: The idea of a solution in search of a problem was true of the reading machine because we had this solution to recognizing ordinary print and we’re looking for what that might be good for. The music synthesizer came about a little differently actually. Our first customer of the reading machine was Stevie Wonder and that started a relationship with him, a friendship in 1976.In 1982 we had a conversation about the state of the art in musical instruments and on the one hand there were these 19th century acoustic instruments that created sounds of choice, beautiful sounds like the piano and the violin.

But they were very hard to play and you couldn’t play them at the same time and most musicians couldn’t play more than one or two instruments and most of them only played one note at a time. And then there was this emerging computer world where you could play a line of music and the computer would remember it. And you could play it back from memory and play another line over it and build up a multilayer composition. But the sounds that you had in that world were very thin, they were basically electrical sounding. If you selected piano it sounded like an organ, and so on.So wouldn’t it be great if we could combine these two worlds of music, have these very powerful control methods of the computer world with these beautiful sounds of the acoustic world.

And so I thought about it and actually felt that pattern recognition could provide a solution. So Stevie Wonder and I actually started Kurzweil music systems in 1982. In 1984 we introduced the Kurzweil 250 which was recognized as the first electronic instrument that could realistically recreate the sounds of orchestral instruments like the grand piano.


Erin: Talk about Stevie Wonder and that collaboration. What is your favorite Stevie Wonder invention composition?

Ray Kurzweil: What is my favorite song of his?

Erin: Yes or you know, based on the work that you’ve done together.

Ray Kurzweil: Well, he did an album, Songs in the Key of Life, on our instrument. He did quite a few albums around that time using the Kurzweil 250 and successive instruments but that’s a particularly beautiful album that I recall. But he has a whole lifetime of wonderful music.


Erin: Indeed. You grew up with music, your father a composer, obvious huge impact on your life- your documentary Transcendent Man evokes the sense of loss- do you want to bring him back?

Ray Kurzweil: Well we will be able to create virtual people. We have avatars today- they’re not at human levels yet. But if you go on Second Life for example, which is a virtual world, you’ll meet other people. Now most of the avatars you meet there have a biological person behind them. But there are actually avatars walking around in Second Life who are just ‘bots’ as they’re called, they’re artificial intelligence. There’s no biological person behind them.

There’s actually a game going on, how long can those bots pretend to be/ have a biological human behind them without people discovering that they’re just artificial intelligences. They’re doing better and better but they’re still not passable for humans.If you go out in the future, go to the 2030s, first of all Second Life will be very realistic. It won’t be cartoon-like, it will be as realistic as real reality. It won’t be a little flat picture on a screen. It will be full immersion. You’ll be inside it. Instead of having an avatar that you look at in front of you you’ll feel like your body is the avatar and it could be a different body than you have in real reality.

Just like in Second Life you’ll meet other people and some of them will be projections of biological people but some of them will be ‘bots’ as they are today. But unlike today they will actually be indistinguishable from biological humans.We’ll be able to create artificial people in the way that I’ve just described who have the personality of other people who maybe happen to be alive or happen not to be alive based on the information we have about that person. So it’s one of the reasons I’ve actually kept all of my father’s papers and he kept all of his papers. I think he had some intuitive sense that that would be valuable. I have all of his letters, all of his music, videos, pictures, his doctoral thesis, even his bills, and you know we can access his DNA from his gravesite.

Future artificial intelligences will be intelligent enough to create an artificial person that passes a Turing Test for that person, a Fredric Kurzweil Turing test. This artificial person will be indistinguishable from him to the people who knew him like myself. Admittedly my memories of him are fading but that’s actually another source of information. These AI’s can go inside my brain and extract those memories. So will that be my father? You can certainly make a strong philosophical case that it’s not, that it’s a new person that just happens to be similar to him. On the other hand if he’d actually lived, and he’d be in his 90’s now, he’d certainly be very different than he was when he died at the age 58. In fact all of us are constantly changing. I’m not the same person I was six months ago. In fact in terms of particles I’m completely different. Every particle is a different one than it was six months ago.


Erin: Talk more about particle changes…

Ray Kurzweil: Well, with that statement I was making a general statement about everyone. All of our cells change, most of the cells actually die within a few days or a few weeks and are recreated. The neurons persist but the parts of the neurons like the tubules and filaments and iron channels, die and are recreated. So six months later you are literally different stuff, you’re not the same particles or even cells that you were six months ago. There is a continuity of pattern so that’s where the identity comes from.That’s why I say that what we really are is a pattern of information. I make a comparison to, of water in a stream. Look at water going around a rock.

That pattern that you look at can be the same for hours or for years but obviously the water that makes up the pattern is completely different. It changes in a matter of a second. We’re kind of like that pattern of water in a stream. There’s an ancient Chinese saying that you can’t walk in the same river twice because you go back to the river and it’s different water. But the pattern of water will be similar. So we are a pattern, that pattern changes slowly but there’s a continuity. In that sense I’m the same as I was six months ago, I at least have a continuous link to that person in terms of pattern of information. But if you think of yourself as no, I’m really this stuff here, this arm, this physical set of particles, that’s not the case. We continually change that pattern.


Erin: Why aren’t we seeing more AI and other major technology spread around the world? I mean, is it relevant to those who live in dire poverty?

Ray Kurzweil: We see a lot of it. When people say, you know, where is all the artificial intelligence it reminds me of people who go into the rainforest and say where are all the species that are supposed to be here? When in fact, there’s 50 species of ants within 25 feet of them. But they don’t see any of them because they are hidden inside the ecostructure.Artificial intelligence is hidden in our modern economic infrastructure but it’s all around us.

Intelligent algorithms fly and land airplanes, they diagnose diseases, blood cell images, electrocardiograms, they guide intelligent weapons systems, automatically detect credit card fraud, help design products, manufacture them, I could list a hundred applications of artificial intelligence. And these were research projects 10 or 15 years ago. And they’re now really relied on in our modern economy. These are doing tasks that used to require human intelligence. And doing them at human levels and beyond.We’re continuing to automate jobs at the bottom of the skill ladder but we’re adding new jobs at the top of the skill ladder.

So more and more technology goes into education and more and more investment goes into education. In the United States, we had 60,000 college students a century ago, we have 6 million today. AI is really all around us and yes it’s getting more and more intelligent and yes it’s all over the world.Fifteen years ago, if someone took out a mobile phone in a movie that was a signal that this person was very wealthy because only a wealthy person could have a mobile phone. And they didn’t work very well.

Today half the world has mobile phones. They’re not just phones, that’s a misnomer, they are very powerful computers and communication devices. You can access all of human knowledge with a few keystrokes. I just got back from China, half the farmers in China have these devices in their pocket and they’re not wealthy. They are ubiquitous in Africa. There are many projects now, in fact, to bootstrap an educational system and health information system. To help diagnose diseases for example, using this ubiquitous platform of computing and communications.


Erin: Singularity University and the movie Singularity Is Near- is that a kind of generation-Y version of Blade Runner? Because as our intern says in his question, Ramona the female alter ego sure looks like a replicant.

Ray Kurzweil: Singularity University actually opens next week on Monday. It was our view that the time was right to start a university like this because information technology now is really emerging to be powerful enough to address major world problems. It’s really only the exponentially growing powers of these information technologies that have the scale to address the major problems of the world such as energy and the environment, health and medicine, longevity, disease, poverty, and so on.

This university is sponsored by Google; it is on the NASA space agency campus in Silicon Valley. It will be permanently headquartered there. We have about 50 faculty who are the leaders in all the areas we’ve been talking about- artificial intelligence, nanotechnology, biotechnology, and the applications of those technologies. We have 40 students who will start next week. After this semester we’ll expand to 120 students. We had 1200 applications for those 40 slots, and these are very exciting students. We are going to be starting a network of people who will have been involved with Singularity University. There will also be a collaborative communication component so people can take these courses and see the lectures and participate remotely through electronic communication.

There are two movies related to my ideas, you mentioned Transcendent Man which is a movie, I didn’t make it, it was made by a very talented filmmaker Barry Ptolemy. It’s about me and my life, career, and ideas. It premiered at the TriBeCa Film Festival and was very successful; it’s a very beautifully made movie. I was very pleased with it.

I’m also making a movie based on my book, The Singularity Is Near, it’s called The Singularity Is Near: A True Story About The Future. It intertwines a documentary thread where I interview 19 big thinkers like Alan Dershowitz or Marvin Minsky or Richard Clark and so on, including Alvin Toffler, the author of Future Shock, about their ideas about the future. Also their ideas about my ideas.That’s intertwined with a B-line narrative story, a fictional story about Ramona. It actually starts in a true past with my creating Ramona in a mechanical puppet theater when I was eight. I demonstrated an electronic version of Ramona at the 2001 Ted conference, that’s in there, and then Ramona goes in the future. She becomes more and more realistic, more and more human-like, more and more independent, kind of a Pinocchio story. She hires Alan Dershowitz, the famous American civil liberties lawyer to press for her legal rights as a person. He plays himself. She gets coaching from Tony Robbins to learn the secret of what it means to be human. And the story goes on from there. And the story is illustrating the ideas in the documentary thread. Both movies will be coming out later this year.


Claire: Your legacy, what do you hope it will be assuming that you pass away, at least temporarily?

Ray Kurzweil: Well, I do hope to be around so that’s one legacy I hope to maintain. But aside from that I think the most important idea that I’m trying to communicate is this exponential growth of human knowledge and of information technology. That ultimately will transform everything -health and medicine is already information technology. That was not the case a few years ago. All of these different areas become information technology, energy is becoming information technology as we’re applying nanotechnology to things like solar panels and exponential growth is ultimately very explosive.

If we look clearly at the implications of that, we see very profound transformations. But it will be a continuation of what we’ve seen already, which is that we’ve always used our tools to expand our reach ever since we picked up a stick to reach a higher branch. We’ve extended our reach with our tools first physically, now mentally. We carry a device in our pockets that can access all of human knowledge with a few keystrokes. That wasn’t possible just a few years ago.Ultimately these devices will go inside our bodies and brains and make ourselves smarter and expand our intellectual and emotional reach. They are also dangerous. Technology is a double-edged sword so I also try and educate people on how we can keep these technologies safe while we maintain the benefits of overcoming human suffering. There’s still a lot of human suffering to overcome.


Claire: When the Singularity occurs, will Artificial Intelligence have the same motives that we would? Would they be curious? And the idea that they would expand out into the universe, would they have an interest in doing that or would they just sit there? Would they care what happens to humans or the planet?

Ray Kurzweil: Well, I think that they will be curious- that is, I think expansion of complexity and knowledge is an inherent part of evolution. We saw that even before humans were around, that biological evolution led to more and more curious creatures, more and more intelligent creatures, and finally to homo sapiens. And then we’ve used our technology to expand that.When you refer to these intelligent AI’s as they, I think they’ll be us. We’re going to merge with that technology, it’s not a species apart.

It’s just not some alien invasion of these intelligent machines from Mars. They are emerging from our civilization. They’re extensions of who we are, so I think they will reflect our values as well as our conflict of values. I’m not predicting that conflicts will go away. This is not a Utopian vision, but you’re not going to be able to walk into a room in 2035 and say okay, well humans on the right side of the room, machines on the left. It’s gonna be all mixed up- a biological human will have artificial intelligence in their bodies and brains. Ultimately that will be where the action is, because that is going to grow exponentially whereas our biological intelligence is going to remain fixed. It is an expansion of who we are and it’s going to be derived from human intelligence. Ultimately it will be very much like us.


Erin: Fascinating, I can’t wait for the future.

Ray Kurzweil: Me too, that’s why I’d like to stick around.

Erin: Thanks Ray. Ray Kurzweil is a leading inventor and futurist. You can learn more about him and his work here: We hope you can come back on Wednesday, July 1 at 1800 utc when we get a preview of a special Independence Day event that will allow visitors to return to the giant crown structure on top of the Statue of Liberty. That’s Wednesday, July 1 at 1800 utc on See you then!