Ray Kurzweil was the principal developer of the first omni-font optical character recognition, the first print-to-speech reading machine for the blind, the first CCD flat-bed scanner and the first commercially marketed large-vocabulary speech recognition. He's a big name in artificial intelligence, nanotechnology, and — what's this?! — advances in extended, healthy lifetimes. Kurzweil has received seven national and international film awards and also received the 1999 National Medal of Technology, the nation's highest honor in technology. He's the author of several books, including The Age of Spiritual Machines, When Computers Exceed Human Intelligence.

In this interview with DevSource, he discusses the role of creativity in software development, the future of programming, and the redefinition of the thinking machine.

DevSource: In The Age of Spiritual Machines, which came out in 1999, you predicted that by 2009 computers would be embedded in clothing, most routine business transactions would take place between a human and a virtual personality, and translating phones would be in common use. Do you think we're halfway there? If not, what's keeping those innovations from arriving (or reaching an "unremarkable" status)?

Ray: Technical progress is exponential, not linear. So 10 years into the 15-year genome project, only a couple percent of the human genome had been collected. But we were doubling the amount of genetic data sequenced each year, and it's the last seven doublings that go from one percent to 100 percent of a project. The project got done in time despite skepticism from "linear" thinking skeptics.

In 1999, notebooks were not yet fully ubiquitous, most PCs were desktops, and hand held devices were very primitive. Today, there is a consolidation of e-mail, Web browsing, hand-held computer functions, PDAs, games, phones, cameras, videorecorders, and music players in remarkably small pocket sized devices. There are prototypes of translating telephones. There are virtual personalities engaging in natural language conversations conducting routine transactions (for example, British Airways' virtual reservations agent). So I believe we're on track. In another five years, the price-performance of computing and communications will improve by another factor of 32 over today.

DevSource: The people reading this interview are the software developers charged with writing the applications we'll be using in the futures you've envisioned. What do you think those developers need to know? What are you afraid they're not thinking about?

Ray: As point to point communication becomes more ubiquitous, reliable, and high bandwidth, we will be migrating to grid computing in which we can readily access the unused computes on the Internet. Each device, whether in a fixed location, a device in a pocket, or embedded in clothing or the environment, will not just be a "spoke" into the Internet, but will be a node passing on messages from other machines. So communications will be constantly reconfiguring itself and self-organizing with every device part of the network. In addition, every device will make its "computes" available in the same way. So there will be less concern with peak computational needs. If you need a supercomputer for a few milliseconds (or more), you'll have instant access to it.

To make this fully effective, however, applications will need to be optimized to run in a massively parallel way. This is how the human brain works (with about a hundred trillion simultaneous processes in the interneuronal connections). So developers need to think how their applications can take advantage of the inherently massively parallel nature of the computations that will be available to them on the Internet.

DevSource: In your writing, you've mentioned that the human tendency to pervasively accept innovations — such as AI and machine intelligence — causes it to become invisible. And, as a result, AI has become "the pursuit of difficult computer science problems that have not yet been solved." That's surely true for my 85-year-old Mom, who isn't quite sure how e-mail works and simply accepts the magic as delivered. Are developers (the people creating tomorrow's innovative solutions, or at least tomorrow's payroll processing) equally blind? Should they be?

Ray: As we master and understand a technique, we think in terms of that technique — Markov models, genetic algorithms, search techniques, signal processing methods — and not generally about "AI." As we progress through the reverse-engineering of the human brain, we will expand our AI tool kit to incorporate the brain's methods for learning, pattern recognition, and decision making.

Brain reverse engineering has not contributed that much to AI to date because we have not until recently had the tools to see the brain in action at sufficient temporal and spatial resolution. Imagine if you were asked to reverse engineer a computer, and all you were given were some crude magnetic sensors. The spatial and temporal resolution of brain scanning has been increasing exponentially, and the latest generation of scanning tools can see individual interneuronal connections operate in real time in substantial clusters of neurons in living brains.

We've also shown that we can turn this data into working models and simulations, for example effective simulations of over a dozen regions of the auditory cortex (Lloyd Watts and his group), and a working simulation of the Cerebellum, which comprises more than half the neurons in the brain. So ultimately, we will understand how the brain works in terms of specific methods.

Most mainstream applications in a wide range of fields incorporate techniques that were AI research projects only a decade ago. Examples include search engines, automated investing, credit card fraud detection, automated analysis of electrocardiograms and blood cell images, monitoring intensive care units, flying and landing airplanes, guiding weapon systems, and many others.

DevSource: You've created a wide range of inventions, and have been involved in everything from OCR to speech recognition to nanotechnology. What — if anything — do they have in common? Is there a central theme to your endeavors? (Surely, many people who think of you as "the guy that invented all those computer things" would be surprised to know your latest book, Fantastic Voyage, is about life extension.)

Ray: My primary area of technical interest and expertise is pattern recognition, which, incidentally, comprises the primary strength of human intelligence. We're not very good at analytical or logical analysis. Computers can already outperform us in those areas. But we retain an edge in recognizing patterns. So most of my major inventions — omni-font optical character recognition, speech synthesis, music synthesis, speech recognition, financial pattern analysis — involve pattern recognition, or at least the study of patterns. Typically we use self-organizing methods, and it is in this area that brain reverse-engineering will be particularly helpful.

I realized that my inventions had to make sense when the project was finished, rather than for the world that existed when the research project began, and invariably the world was a different place three or four years later. Most technology projects fail not because the inventors are unable to get the thing to work, but because the timing is wrong. So I became an ardent student of technology trends. I now have a group of ten people assisting me to gather data on key trends, and we develop mathematical models of technology evolution that have proven remarkably accurate over the last couple of decades. So this enables me to essentially invent with the technologies of the future.

While I can't build a device using computers circa 2020 today, I can envision what they will be capable of doing. A lot of my technology writing is driven by this approach.

I came to health through my own health issues (cured my own type II diabetes twenty years ago after the conventional treatment proved counterproductive, and wrote a best-selling health book about this in the early 1990s). Today, my ongoing interest in health issues has now merged with my interest in technology because we are in the early stages of the biotechnology revolution in which we are learning to understand, and to control, the information processes underlying our biology.

DevSource: With such a long list of achievements, it's easy to wonder what you do differently. Do you think you approach problem-solving with a different perspective? Is it an ability to see opportunity in a conversation with Stevie Wonder about what synthesized music can't do? Or do you get charged up to solve an unsolved problem? (Or is the pursuit of "where does creativity come from?" too much like pulling up the turnips to see how they're growing?)

Ray: This is a complex question, but I'll share one method I use on a regular basis. Each night, as I go to bed, I assign myself a problem. It can be anything from a math or technology problem to a business or interpersonal issue. I try not to solve it, but to review what I know about the question, and what form an answer might take. I then go to sleep and will find that I dream about the issue often in strange ways.

In our dreams, the "censors" in our brains are relaxed. That's why we often dream of things that are socially taboo. But we also relax our professional censors, the parts of our brain that say "you can't solve a signal processing problem in that way." So new and innovative ideas get tried out. But something else that is suppressed in dream state is our rational function, which is needed to evaluate ideas.

Not every new idea is a good one, most of course are not. That's why in a dream strange things happen and we don't consider them strange, at least not while dreaming. So then, in the morning, in a state halfway between dreaming and waking — lucid dreaming — the dream state continues but I now have enough rational awareness to evaluate ideas critically. I now return to the problem I assigned myself, and invariably I will have important new insights.

DevSource: I'm sure people ask you, often, "Where do you get your ideas?" Why do you think they ask that?

Ray: One source is to think methodically of where technology is going and what will be feasible when. So often I have a project in mind but realize the enabling factors are not yet in place. Once I feel that they will be in place around the time that a project could be completed, and there appears to be a substantial need, then I will consider starting that project.

DevSource: What are you most proud of?

Ray: The exciting thing about inventing is the link between dry formulas on a blackboard, and transformations in people's lives. So hearing from blind persons who say that the reading machine enabled them to complete their education, or to get albums from musicians who were able to create new types of music with our synthesizers, that's a thrill.