10/3/97

CONTACT: Eleanor Selfridge-Field, CCARH (650) 725-9242
e-mail: esf@ccrma.stanford.edu
David F. Salisbury, News Service (650) 725-1944
e-mail: david.salisbury@stanford.edu

Frontiers of computer intelligence subject of symposia

Computer scientist and Pulitzer Prize-winning author Douglas Hofstadter has organized a series of public symposia designed to probe the provocative question, "Are computers approaching human-level creativity?"

Hofstadter, a professor of cognitive science at the University of Indiana who is currently a visiting scholar at the Stanford-based Center for Computer-Assisted Research in the Humanities, has been a vocal critic of a number of major projects in the field of artificial intelligence. In his study of the field, however, he has discovered several different areas in which computer scientists have written programs that produce amazingly human-like results.

The first four symposia explore what Hofstadter considers the most interesting of these areas: chess and the game go; language and literature; musical composition; and puns and humor. Proponents and prominent critics of each of these developments will discuss the extent to which the computer programs involved are reproducing human thought processes in a profound fashion or merely using various techniques and tricks that can produce similar results.

• The most familiar of these topics is computer chess. Programs for playing chess and go is the topic for the first symposium, which will be held on Saturday, Oct. 11. Participants include Eliot Hearst, a Columbia University psychologist who is a former captain of the U.S. Chess Olympic Team and has written critically about computer chess; and Monroe Newborn, a computer scientist from McGill University who developed one of the early chess-playing programs and argues that programs like Deep Blue are "thinking" when they play chess. The session also includes a presentation by Tim Klinger, a New York University graduate student who has spent seven years developing a sophisticated go-playing program. The symposium will be held in the Stanford Law School, Room 290, from 10 a.m.-3 p.m.

• Language and literature is the subject of the second symposium, which will be held on Saturday, Oct. 18. The focus of the discussion is on the program ISAAC, which its developer, Ashwin Ram, a computer scientist from Georgia Tech, argues can "understand" works of science fiction. Facing off with Ram will be San Diego cognitive scientist Gilles Fauconnier, who studies the cognitive underpinnings of language, and Mark Turner, a Maryland University English professor and expert on the relationship between cognitive science and literature. The meeting will be held in the Stanford Law School, Room 290, from 10 a.m. to 3 p.m.

• The third and largest symposium, which will be held on Saturday and Sunday, Nov. 8-9, deals with musical composition. It centers on two programming efforts: a musical-style-imitation program called EMI, developed by music professor David Cope at the University of California-Santa Cruz, and the melody simulations of music professor and jazz improviser Steve Larson at the University of Oregon. According to Hofstadter, Larson's treatment is more realistic psychologically, but Cope's efforts have produced the most life-like compositions. Tufts University philosopher Daniel Dennett, Boston programmer and musician Bernard Greenberg, Hofstadter and musicologist Eleanor Selfridge-Field from the Center for Computer-Assisted Research in the Humanities will provide differing perspectives on the subject. The meeting will be held in Campbell Hall in Stanford's Braun Music Building and will run from 11 a.m. to 5 p.m. on Saturday and 10 a.m. to 4 p.m. on Sunday.

• The fourth symposium, which is scheduled for Nov. 21, deals with the creation of puns and humor. It will feature Kim Binsted ­ a cognitive scientist from the University of Edinburgh who has developed Jape, a program that concocts original puns in several languages ­ and comedian and actor Steve Martin.