July 24, 2008
So, following up yesterday's rant about the Chinese Room...
Reading further into the book, I see what Hawkins is up to. Around 100 pages in he writes
If Searle's Chinese Room contained a similar memory system that could make predictions about what Chinese characters would appear next and what would happen next in the story, we could say with confidence that the room understood Chinese and understood the story. We now see where Alan Turing went wrong. Prediction, not behavior, is the proof of intelligence.So now we see where Hawkins went wrong... Turing specified a judge looking to determine if the conversation partner is a human or a computer, and is permitted to ask questions that could not be answered without having a normal human's ability to predict the flow of a conversation, to fill in the gaps. Thus Hawkins use of the Chinese Room is a giant strawman, where he might be using the room as a stand in for "computers as they are generally used now" (with a CPU, long and short term memory, following programs step by step, etc) and a weak form of the Turing test (fooling a Chinese speaker who probably wasn't having that deep of a conversation to begin with) and saying that this test can be passed by a machine that isn't really thinking, which is view so weak it's tough to argue against.
For Hawkins, and I think he makes a strong case for this, prediction - a non-stop giant flow of expectation and comparison with reality - is the tool and hallmark and perhaps even necessary component of intelligence. He is probably taking for granted Searle's idea of "Strong AI" vs "Weak AI"; some proponents of the former would argue that even a simple thermostat has a (extremely) rough form of consciousness, that it in effect "wants" the room to be a certain temperature and "acts" according to that desire. Hawkins sees a bigger, unbridgeable gap between that kind of simple mechanism and generalized intelligence, rather than a continuum, and feels that he has isolated the crucial difference.
I like when I read a book about how the brain and consciousness might function, and suddenly I feel more self-aware of my own internal thought process.
Quote of the Moment
"Sure it mattered. When you get to my age you discover that everything mattered. Life isn't a series of good and bad choices. It's harder to steer it one way or the other than most people think. You just get pulled along. You look back and you wonder 'could I have changed the course of my life?' Maybe you could've ... but it would probably have taken a tremendous force of will."
--Old Man in Seth's "It's a Good Life, If You Don't Weaken", a graphic novel I just read. The man was a friend of "Kalo", a New Yorker cartoonist the main semi-autobiographical main character is trying to find information about. (It turns out Kalo is made up by Seth (pen name of Gregory Gallant), though he throws in some convincing mockups of Kalo cartoons at the end that really make the quest feel real.)
I think I should let myself be hungry more -- "full" as default is probably not good. Conversely... diet coke and creme de menthe altoids?