diana@csri.toronto.edu (Diana Li) (07/18/89)
FLASH ANNOUNCEMENT
(SF = Sandford Fleming Building, 10 King's College Road)
-------------------------------------------------------------
AI SEMINAR
SF 1101, at 11:00 a.m., Thursday 27 July 1989
Henry Kautz
AT & T Bell Laboratories
"Hard Problems in Soft AI"
I'll discuss the role of special-purpose inference mechanisms in
commonsense reasoning. The title "Hard Problems in Soft AI", has a triple
meaning. The first involves methodology: I'll talk about some work by
myself and others which attempts to pluck crisp, well-defined problems from
the softer, murkier areas of AI. The second meaning refers to some
specific technical results we've uncovered, which show that some seemingly
elementary kinds of reasoning are quite hard to perform in general.
Finally the title refers to the problem of integrating the various
specialized inference systems into a coherent general architecture for
intelligence. This problem of synthesis is a crucial challenge to the
symbolic processing approach to AI.