(All photos by Asa Mathat)
Founder, Kurzweil Technologies
Kurzweil: In the ’70s, I was running my first company, major company. We invented the flat-bit scanner and OCR. And I had this vision: “I’m going to build this up to be Xerox,” which was the Apple of its day. And my CFO convinced me that no, that’s actually not what you are good at. You are good at coming up with these inventions and building them up to a certain point. So instead we sold it to Xerox, and then that worked out very well. It was a good deal. Xerox spun it out. Today it’s Nuance, worth about $10 billion. And actually it’s worth about a quarter of what Xerox is worth. So didn’t quite get up that big, but, okay.
Right around the same time, I started to realize that timing was critical to being successful as an inventor. Turns out to be important for everything, you know, from IPOs to romance. So I started to actually study this methodically. And I shared the common wisdom that you cannot predict the future—maybe I could get some educated guesses—and made this surprising discovery.
Not everything is predictable, but the fundamental measures of information technology follow these amazingly predictable trajectories through thick and thin, through war and peace. The measure of computing, calculations per second, per constant dollar was a very smooth exponential curve going back to the 1890 Census. I then laid that out through 2050, continued that curve. We’re now in 2012 and we’re right where we should be. It continues to surprise me as new data comes in and follows these predictable curves. Because what we’re measuring is actually innovation and creativity and competition. It turns out to be predictable.
And finally, I mentioned yesterday what I have been thinking about thinking for 50 years. And up until a few years ago I did share the common wisdom in the neuroscience field that we have these different specialized regions that do different things. V1 is good at edges and very simple shapes, and fusiform gyrus is good at faces, and the frontal cortex is famous for being able to master language. And I figured they must be structured very differently. The one that does these high-level concepts must be very sophisticated and very recent in an evolutionary basis. What’s really surprising is that they are all the same. In fact, research that came out just as I released my book shows that they are interchangeable. In a congenitally blind person, V1, which has these very primitive visual image recognitions, can actually do high-level concept recognition. The only difference between them is where they stand in the hierarchy of these pattern recognizers in the neocortex. And the hierarchy is something that we create ourselves with our thoughts. So from the moment we’re born, or even earlier, we are creating this grand hierarchy that defines who we are.