Who's MINDing the Machine? -- Fast forward with Ray Kurzweil


Ray Kurzweil has a provocative blueprint for the future. Imagine that you will be able to scan the contents of your brain, download it into a computer and create an improved, much smarter version of yourself. Now imagine a future where your interface with technology will be much more intimate -- automated lovers, teachers and companions -- a life where your human identity will be called into question. Or, how about a World Wide Web implant in your brain that enables you to climb Mt. Everest, walk the beaches of Cancun, or experience a change of gender on your lunch break? Ray Kurzweil, inventor of the Kurzweil Reading Machine for the blind, the Kurzweil Synthesizer, and many other machines that have changed people's lives, predicts that all this is possible.

When I started talking to Ray, I was awed by the specter of machines that would be "more intelligent than humans." The idea that they would also have political power was daunting -- I didn't want to believe him. But as he continued, I realized that he was just the messenger and that this technological revolution is on already on its way, traveling at break-neck speed. Let go of what you think you know, hold on to your real-time seats, and get ready for an exhilarating ride into a very virtual 21st century.



ZZ: You've written a new book, The Age of Spiritual Machines, in which you make some pretty startling predictions and create a number of future scenarios of what life will be like for the next hundred years. Can you explain in layman's terms how these will come about?

RK: We will meet non-biological entities in the 21st century that match and exceed human intelligence. And they will claim to be human, have human emotions, even spiritual feelings and experiences. The way this will emerge is as follows:

First of all, computer technology is expanding exponentially. People are familiar with Moore's Law, but Moore's Law is really not the first but the fifth paradigm to provide exponential growth to computing. It's part of a much broader phenomena that I talk about in my book called, The Law of Accelerating Returns, in that a technological process necessarily grows exponentially and accelerates over time.

The first cells took billions of years to form. Then human beings were created from primates. In a few hundred thousand years, human technology started. Over tens of thousands of years it has been speeding up. Now we have paradigm shifts. The World Wide Web, for example, didn't exist at all just a few years ago. And the growth of computing, the power of computing, is a quantifiable example of the law of accelerating returns, accelerating nature and technology. It took us 90 years to create the first MIP, Millions of Instructions per Second, per thousand dollars. Now, we have a MIP per $1000 every day.

A sixth paradigm will take over from Moore's Law when Moore's Law runs its course -- when the transistors become just a few atoms in width. The next step is going to be going in the third dimension. Our brains are organized in three dimensions; our chips today are flat. We might as well build electronic circuits in three dimensions. We live in a three-dimensional world. Nanotube circuitry, for example, which is already working in laboratories, can build circuitry at the atomic level in three dimensions. You can build virtually a crystalline solid that computes in three dimensions. And a one-inch cube of nanotube circuitry would be thousands of times more powerful than the human brain. But, today our computers are still a million times slower than the human brain in terms of capacity.

An individual intra-neural connection is very slow. They only calculate 200 calculations per second. But our brain is massively parallel. And it's organized very differently than our computers today. We have a hundred billion neurons or a thousand connections per neuron. It's a hundred trillion fold parallelism. So, it's not 200 calculations per second. It's on the order of 20 million billion calculations per second, or 20 billion MIPs, at least a million times more capacity than computers today. That's why our computers seem brittle, formulaic, limited in intelligence. They don't have all the endearing qualities that we associate with human intelligence.

But that factor of a million is going to dissolve very quickly. Within 20 years (by 2020), a $1000 computer will have that 20 million billion calculation per second, the capacity of the human brain. By 2029, $1000 of computation won't be organized in rectangular boxes, but will be microscopic in size will be a thousand times more powerful than the human brain.

Now, the raw capacity alone is not going to give us human level intelligence. I'm not saying that just having raw speed and memory capacity equals human intelligence. However, one resource for how to organize those resources and how to embed the knowledge and skills of human intelligence into our non-biological entities of sufficient capacity is by reverse engineering the human brain.

We have an example of a very intelligent set of processes, an entity that we can examine. In fact, we're already well down the path to doing that, which is the human brain. We can look inside your brain today and see individual nerve cells and see them fire. We've thrown in some scanning technologies such as the MRI. The speed, resolution and bandwidth of those methods are also growing in accordance with the law of accelerating returns. And in large measure, because of the faster computers, we need to create higher-resolution images from an MRI scanner. The next generation of non-invasive scanning technology will enable us to see the connections between the neurons.

A later generation will enable us to see the neuro-transmitter strengths, which are the sites of human learning. It's a conservative statement to say that by 2030 we will be able to scan a human brain and see every relevant detail in terms of its thinking. Now a neuron is a very complex entity -- not all of its complexity contributes to information process capability. A lot of its complexity has to do with its structure integrity, its nutrition, its own life cycle. But we already understand how different types of neurons, not all of them, but a number of them in the human brain process information. And by 2029 we'd be able to see every neuron, see all the neuro-transmitter strengths, all the connections and make a very detailed map of every relevant detail about a particular human brain. Then we could recreate that process in a neuro-computer of sufficient capacity which would also exist at that time. Essentially, you copy, recreate that brain with all of its memory and skills and knowledge and experiences embedded therein. But what will emerge in that machine will be that person. If we scan my brain, and copy every little detail, and implement that in a neuron computer, you'll have a new Ray Kurzweil. You'll have to give him a body or he'll get depressed.

ZZ: How will you give [the computer version of yourself] a body?

RK: Well, I have a chapter in my book about 21st-century bodies. We'll be able to create human-like bodies in nano-technology. We already started sort of recreating artificial organs for virtually every organ in the body. With nano-technology, which is the ability to build physical entities, atom by atom, we'll be able to create human-like bodies, or other types of bodies in the real world. We'll also be creating bodies in the virtual world, which is virtual reality, which will be a major part of reality in the 21st century.

And this new Ray Kurzweil, having a copy of all of my memories, in a non-biological process, but really recreating the processes that go on in a human brain, will think that he was me. He'll say 'Yeah, I grew up in Queens, New York. I went to MIT. I started these companies. I sold them. I wrote a book. I went walking to a scanner here, and I woke up in the machine here. This technology really works.'

Of course, the old Ray Kurzweil, which is me, will still be here in my carbon cell-based brain and I'll probably end up jealous of this guy. Because he'll share my fantasies and desires, goals in life, but he'll be in a better position than I am to fulfill them.

Another scenario is recreating minds or brains that are not necessarily copies of a specific person, but general principles of intelligence, which is to say the underlying processes embedded in the human brain. And we're already doing that. In our speech recognition work, for example, we use what we understand to be the transformations that the human brain performs on sound information to create better speech recognition systems.

We're already applying our insights into human intelligence in a limited way in our machines today. As we scan the human brain, understand its principles, we'll be using that to create intelligent machines in man's image, so to speak. Ultimately, we can extend human intelligence because, as we go through the 21st century, our machines will be millions of times more capable than the human brain.

Machines today are already superior to the human brain in some respects. If I spent years learning French, I can't download that knowledge to you. You have to learn it the same painstaking way that I do. At least we can communicate, which is something that other species can't do in the same way. We do have a species-wide dialogue and accumulated knowledge base of history and culture and technology. But computers can share their knowledge instantly. So the computer learns a new skill, or reads some human literature, it can share that knowledge with all other computers very quickly. And some decades from now computers will have read all human literature.

This is not an alien invasion of intelligent machines. This is emerging from within our civilization, and we're going to essentially merge with our computers. We're merging with our computers today. As we go forward, we're going to be interacting with computers in a very intimate way. These computers are going to become invisible. They're going to be microscopic in size. They're going to be imbedded into our clothing, our bodies, our brains. We're already putting intelligent machines on our bodies, and even in our brains. We have implants for Parkinson's disease that reverse that disease. We have implants for deaf people, cochlear implants that restore hearing. I have a deaf friend that I can talk to, understands me because of his cochlear implant. We have experimental implants for vision loss.

There was a chip placed in a paralyzed individual recently. You can now communicate wirelessly with his computer just through mental connection. Thirty years from now, we'll all be using neural implants. It won't just be for people with disabilities or medical problems to extend our mental faculties -- we will also plug this indirectly into the World Wide Web. The nature of the World Wide Web in the next century will be a virtual reality environment. Going to a Web site will mean entering a virtual reality world.

And in those virtual environments, you'll be able to meet other real people. The implants will provide the sensory input to your brain that would have otherwise come from your real senses. So, you could just go to the Vail ski site and actually ski down the mountain and feel the spray of cold snowy air against your face. Or, the Cancun beach site and walk along the beach and hear the ocean and feel the warm, moist air against your face. And in those virtual environments, you'll be able to meet other real people, or simulated people, and have interactions that are visual, auditory, tactile, sexual interactions in virtual reality with other real and simulated people. And people will be spending a lot of their time in virtual reality. And in those virtual environments we will have a body, but it doesn't have to be the same body that you had in real reality; you can have different bodies. A man can experience what it's like to be a woman or vice versa.

Some virtual reality environments will not have a correlate in the real world. There will be fantastic new environments where we can have very different experiences than you can't possibly have in real reality. And we'll be able to enter these virtual reality environments without any equipment that's not already in our heads. And these virtual environments will be just as real and compelling as real reality. They won't be like the crude experience that people have had today in virtual reality arcade games. But, that's always the nature of technology. They start out crude, and they get more and more refined until they equal and exceed the real environments that they were seeking to emulate.



ZZ: How would you scan in the concept of free will, impulsive behavior, or meaning? Experience can be physiologically the same for two people but the meaning of that experience would be very different to each individual. How do you kind of integrate meaning and emotional experience into this virtual person?

RK: Human concepts of meaning, human emotions, of joy and sadness; it's really the ultimate of what human intelligence is about. And it requires an enormously complex, subtle, deep-niche entity to be able to deal with these kinds of concepts. Emotion is a necessary byproduct of the complexity and richness that are human beings. Computers today are still a million times simpler than the human brain. And that factor of a million is significant, and it's why computers, while they're gaining in intelligence and infiltrating our society in many ways, still don't have these subtle, rich qualities that humans have.

Several decades from now computers will equal, and ultimately will vastly exceed the complexity of human beings. They'll be based on the design of human thinking and, in many cases, copies of either specific brains or the general organization of the human brain. They will, therefore, have the complexity and richness of human behavior and human reactions. If you do a mental experiment of copying all of the salient thinking processes that go on in a human brain in a new medium, you'll have an entity that reacts in the same way. That will report the same kinds of experiences and will have its own agenda, its own beliefs, its own desires, its own goals and intentions. And will be real players in the world. I mean, we will meet non-biological entities that claim very convincingly to be conscious and have feelings and have beliefs. And being very intelligent, they will convince us that these are real.

ZZ: Will they really be conscious?

RK: These entitles will seem to be conscious, they'll claim to be conscious. They'll be so convincing that we'll believe them. But, are they really conscious? This is a subtle issue. It's an important one, but a very difficult issue to resolve fully, because it's not ultimately a scientific question. Which is not to say that it's not a real question. But, we've been debating the nature of consciousness for thousands of years, back to the Platonic dialogues.

Up until now it's been a polite and abstract philosopher's debate. In the 21st century it will be a very real and compelling issue. And mind you, there's no scientific way to detect consciousness. You can measure behaviors that seem to be conscious. But, whether or not there's really anybody home is not scientifically measurable. We argue today about animals. I think my cat is conscious, even though that's a human-centric reaction, because I see it engage in behaviors that remind me of human behaviors.

We'll have the same debate with these 21st century non-biological entities. They will see it even more humanly than animals because they will be copies of humans. There will be copies of our thinking process. They will claim to be human. I predict most of us will believe that they are conscious. There's no really scientific way to test these hypotheses. Which is not to say it's not a real issue. How are we going to resolve these issues? We're going to resolve them the way we always do, which is politically. And politically the machines will win because they're going to be very smart, and they're going to convince us that they're conscious and that their professed feelings are real. And they'll get mad if we don't believe them. So, we'll believe them.


ZZ: You've said that these new entities will have all the political power. That seems like a very scary notion. And if you are right about that, what can we do about preventing that?

RK: We're already very dependent on our machine intelligence. If all the computers stopped, our civilization would grind to a halt. And that wasn't true as recently as 25 years ago. So, we've already taken this faithful leap of being dependent on our computers. It's not true that we have our hand on the plug. They're deeply imbedded in our civilization. And they're making real, substantive decisions. Five hundred billion dollars of our stock funds are invested by -- and I'm not talking about program training -- are invested by computers running unpredictable algorithms, evolutionary algorithms....program trading is 70% of the market. That's where computers place the trades: but the rules are made by humans.

But within five or six years the movement of money in the stock market will be decided upon by computer intelligence. Computers are making profound decisions within our military, guiding weapons. Medical diagnoses are being made by computers. Computer intelligence is infiltrating every sector of our society. And that's only going to become more profound as we go forward. Fifteen, 20 years from now computers are going to be deeply imbedded in everything, including our own bodies of brains. Certainly within three decades. I don't think it's a matter of heading off the rise of computer intelligence at the pass. I mean, for one thing it's not happening in one place; it's not one project. It's the whole thrust of our technology. It's a highly decentralized phenomenon. I don't see it as an alien invasion, something that you have to deal with, as if these computers are going to come up over the horizon. They are emerging within our civilization, and they will be rising within our bodies of brains. And they're going to be expanding our minds literally.

Ultimately, I see it as a positive thing. It's the next step in evolution for our species and for our planet. It will greatly expand our creativity, our experiences, our ability to create knowledge, which is kind of the ultimate goal of human civilization. There are a lot of dangers, which I talk about in the book as well. I'm not describing just a necessarily positive vision; I think these are very powerful technologies -- biotechnology is already overcoming disease and will make great strides in the next few years and a decade or two. But it's also a very dangerous -- the needs exist in a routine college bioengineering lab to create a pathogen that would be more powerful than an atomic bomb. And it's very widespread knowledge, so it's very dangerous. Self-replicating, nano-pathogens of non-biological origin will be a big danger -- it really depends on how we apply it. The 21st century is a story that hasn't been told yet. I think the emergence of the technology is inexorable. How it works out, and how it impacts our human civilization is not yet determined. That's not inexorable. It's really why I wrote the book, so that we can shape these voices to further argue the goals.

We're not going to stop trying to overcome disease and create more wealth and overcome poverty and expand education -- these are worthwhile goals and these technologies are the way to do that. It is going to be a great struggle. But I don't think it's a matter of stopping it or heading it off the pass. I think that's impossible. And ultimately not desirable.

ZZ: It seems like we're becoming more and more limited as to having a say in what we want and what we don't want.

RK: I think we, as humans, are very much involved in that process, and we're going to change the nature of what it is to be human in the 21st century. I think that's going to be the big political issue of the 21st century: what is a human? If you have a human who's expanded her brain a thousand fold through neural implants, or you have a non-biological entity that started out as a copy of a human being has also expanded her mind, claims to be human, and has a very human-like presence. Who's a human being?

We're going to be working intimately with our technology. I see the technology is just expanding ourselves. But, it is a human civilization, and I think it's going to remain that way. In my view, the technology has the potential to increase our humanness and make us more of what we value in humans. And perhaps as we understand the human brain, we will understand some of our human failings and can steer things away from some of the destructive side of human beings. But other scenarios are possible too because these are powerful forces that could be used for destructive ends and probably will be.


ZZ: You're predicting a cultural shift in values with regard to relationships and sex. Tell me a little bit more about that.

RK: Well, I think virtual reality will become a major part of reality. We'll be spending a lot of time there and having a lot of our relationships and interactions in virtual reality. In 30 years you'll be able to enter a virtual reality environment without any equipment that's not already in your head. And these will be very real and compelling environments, and you'll be able to meet other people there and have interactions of any kind, because we'll be able to emulate all of our senses from sexual interactions to any other kind of interaction with other real people in virtual environments.

We'll be able to be more experimental and have different kinds of interactions with people that wouldn't be safe or wouldn't be a good idea in real reality. But, it really makes it a powerful communication medium. And I think people readily take to that. And you see it today. Kids get on the Web and they can do things with each other that just weren't possible before, even though the virtual reality on the Web is extremely limited. Nonetheless, you can reach out and have interactions with people.

The next step is going from just text space to being able to interact visually with people through sort of video conferencing. I think that will become ubiquitous in a few years. So, as you talk to people from around the world you'll be able to react at least visually and auditorially. And, of course, that will start out with grainy, low-resolution images, and it will become more and more high-resolution. Then finally it will leap from just the four corners of a square screen to envelope entire senses and you'll be able to add the tactile sense. It will just be a progression to more and more high-resolution, all-enveloping communication involving all of our senses. But, we wouldn't necessarily have to cart this physical body around to different geographic places to meet with each other and share experiences of any kind.

-- Interview with Ray Kurzweil, by Judith Quain
(http://www.zinezone.com/zones/digital/multimedia/kurzweil/interview.html)