Kurzweil

The Kurzweil trifecta is complete. I have now shaken his hand and directly asked him a question. I have gazed upon his leathery skin, and breathed the same air as him.

technolope and I got to the auditorium at around 3:50 for the 4:30 show. It was mostly empty, so we scored seats in the front-est row we could get. The front two rows were covered with little “RESERVED” signs. We made chit-chat with the nice MIT people around us, and generally revved into full-on fanboy mode until the festivities started.

The debate itself wasn’t really all that much of a debate. There was a brief skirmish over who had the burden of proof … and another over what they meant by “intelligent,” “conscious,” and so on. Sort of boilerplate stuff. Then they demonstrated that they each had a mastery of the standard thought experiments. Particularly the Chinese box (in which a man hidden in a box uses a set of rules to shuffle characters around without an understanding of what they mean … but from the outside the symbols are carrying on a conversation in Chinese.) Broadly, they agreed on what was going to be possible with simulation and computation in the next couple of decades … and they agreed on the fact that we were going to see remarkably powerful machines.

One interesting bit was that the original statement of the Turing test specifically states that the machine’s goal is to lie … to fool the interviewer into believing that it is human … Turing elegantly skates around the question of whether that would mean anything about a new “node of consciousness” (or whatever) being created.

One fun exchange went like this:

“You can simulate photosynthesis, but no photosynthesis occurs. You can simulate a rainstorm, but nobody gets wet.”

“Yes, but if you simulate creativity, the ideas that you get will be quite real.”

I went with my gut at the end and bounced up to a microphone to ask a question. I had a little wave of stage fright (sort of refreshing that it can still happen!) at opening my mouth in front of a room full of MIT CS and AI people. Then it passed … but I was still second in line when they declared “last call” on the questions and ended the session.

So I bounced down to the front of the room and joined the gathering fanboy crowd waiting to talk to Kurzweil. He’s got a warm, firm handshake … not crushing, and he seems quite present and engaged when he talks to people. I asked my question, which was about the boundary of the brain simulation he’s so interested in. Right now, there’s an IBM machine being used to simulate a bunch of neurons. Simply by applying the benchmarks taken on this simulation … it’s likely that we’ll be able to simulate a whole human brain, in real time, to the level of chemical gradients, in 2029. My question is, if we simulate the brain, we clearly have to provide inputs to that brain … so we need to simulate a body to some degree. Probably that body needs a world to live in … so we need some level of that as well. My question was, at what point can we stop with the simulations and get on with interacting with this AI? His answer was, in effect, that we’ll meet the beasts halfway. Immersive virtual environments will develop in parallel with these simulations, and he suspects that we’ll meet our AI creations in a virtual middle ground at first.

I also had a fun chat with the other guy … and a grad student who was chatting with him. For this one, my recent buddhist readings came in handy … since the grad student was adamant that consciousness and the 1st person perspective was totally irrelevant. The snitty one-liner response was “Would you rather have surgery with, or without anesthesia?” I made a bit more progress with her by suggesting that the reason acknowledging consciousness matters is ethical, rather than scientific. That the only real source of right and wrong is an idea that my actions have an effect on someone else … that other must be real in order for my ethics to matter.

So, it was fun. The webcast is online (Real format, free player required to play non-open protocol). If you watch closely, you can see when I pop up to ask my question. It’s near the end of the debate.

Then we met up with redmed and watched them set the pagan tree alight in the Commons. Then sushi and sleep.