Ray Kurzweil talk at Intel, Santa Clara
I attended a talk today given by Ray Kurzweil. I started to tweet about it but soon found that the time I took to correct typos in the tweets was detracting from actually listening to Kurzweil speak. I switched over to using Notes on my iPod.
He spent a lot of time upfront telling us why his predictions seem crazy and why we should believe him. To start with, our innate ability to predict something is based on a linear function. This served us well when we saw a predator in the distance and had to figure out how long it would take for it to get to us. This is fine in the short term but hinders our ability to think or design for the future. Second, many critics use limitations in existing technologies as counterpoints. For example, many say that the age of Moore's Law has come to an end. We can not increase the number of transistors per area anymore so the exponential progress is slowing down. To these critics he says, there was a time when vacuum tubes were not going to be any smaller then along came the transistor, and the curve kept on going as predicted. The take-away for me was, his predictions are not about the physical manifestation of the technology, but about the raw capabilities of the technology.
The second major point of his talk was that as technologies are subsumed by information technologies, they to will be subject to exponential gains. His poster child for this point is the health sciences. Before the use of simulations, drug discovery was more hit and miss. New drugs were manufactured and then tested on subjects which created a lengthy process. With advances in drug interaction simulation, researchers can now model a drug molecule and test many possible outcomes before moving to lengthy clinical trials. He then proceeded to discuss the implications of advances in health to increase the capabilities of the human body and the increase of the human lifespan. His basic premise was that our genes are "outdated software" and in the future, we will be able to re-engineer various aspects of our bodies. One example he gave is some recent research in artificial red blood cells. One such cell, is capable of increasing the ability of our blood to store oxygen. Imagine being able to stay underwater, without any oxygen storage apparatus other than yourself for an hour.
Throughout his presentation he showed many graphs detailing trends and how all of the curves transcend the technologies of their time. The curve that stood out the most to me was the progress of raw computational power. The curve showing present developments weren't the interesting parts, what was interesting was that the evolution of organic computational power also fit onto the curve. Our next steps in AI are to combine the incredible processing power offered by inorganic technologies with the right "software" unleashing the next age of intelligent organisms.
In the end, my main take-aways are:
- Exponential growth in technology is something independent of the physical manifestation of technology.
- As other technologies are subsumed, become or are enhanced by information technology, they too will be subject to the exponential gains.
- The exponential growth in computational power not only matches the progress of inorganic computation but is an extension of millions of years of evolutionary progress.