AI & rum
Friday, October 20 2000
After yet another unnecessarily stressful and unproductive day at work, I came home and immediately started drinking rum and cokes with my housemate John, on his suggestion. "Drinking is sort of like drugs to me," John said, explaining, "They're easy to do if they're around." "And it's awfully easy with a fully-stocked wet bar," I agreed. We've had a fully-stocked wet bar since our little house party of a week ago.
We soon found ourselves wrapped up in a fascinating discussion about artificial intelligence and how it relates to relational databases. I told him my big realization of the week, which is "artificial understanding of English text is really just a difficult database modeling problem." I went on to explain that all of us, even the fairly dim-witted, have, in their heads, some sort of database model for English that allows them to make sense of a stream of words. This biological database is full of rules and relationships, associations, images and logic, some of it hard-wired but much of it picked up (populated) by example. Somehow John got to talking about a wacky logic professor he had back in college who had a theory that "truth" is actually something called a "pro-sentence," (like a pro-noun, but a substitute for an entire sentence). This professor had written a book called The Pro-Sentential Nature of Truth. This got me to thinking of other black-box object-oriented language concepts, such as the "pro-Russian Novel."
John went out to see a movie with his friend Chun, and I stayed home, determined to stay awake until he got back so we could go to our coffee shop together. I put Slayer's Undisputed Attitude on the CD player and cranked it up and threw open the doors. But I was so drunk by this point that all I really wanted to do was lie down. Soon enough I was asleep, with all my clothes still on.
For linking purposes this article's URL is:feedback
previous | next