Tuesday, October 31, 2006

all things are not always are

As mentioned in Wired, the standard performance gauge of new electronic devices used to be whether or not it could run Doom. Maybe someday the test will be, can it beat a human in chess?

Whenever I feel like thinking too hard and confusing myself into stopping, I read about artificial intelligence and the research projects associated with it. As technology allows us to scale things down to a human level, the future is not in room-sized supercomputers that can crush master chess players with brute force. The scary A.I. of the future would do that in a $10 handheld device bought at Toys-R-Us.

A few years ago, the Alice Bot (a language-processing chatbot) won the Loebner Prize for demonstrating the most humanlike A.I. ability. I, Robot-style conspirators need not worry, though, as the mechanistic aspects were blown wide open when the bot was set up to chat with itself. The results are quite funny.

Item: Left Behind at the Fishbowl deserves a plug. It's a great cross-section of (generally incompetent) student activity.

No comments: