************************************************************** * * * CYBERSPACE * * A biweekly column on net culture appearing * * in the Toronto Sunday Sun * * * * Copyright 1999 Karl Mamer * * Free for online distribution * * All Rights Reserved * * Direct comments and questions to: * * <kamamer@yahoo.com> * * * ************************************************************** The recent trouncing of world chess champion Gary Kasparov by an IBM computer has chess masters and people in other "thinking" professions contemplating future job prospects. While it's easy to design a phone system that can replace ten switchboard operators, until now, the engineers who have done the designing and other thought workers felt pretty secure. All is not lost. Kasparov accused IBM of rather rudely designing a chess playing machine that can beat him, not a "generalist" that can beat any future world chess champion. It may be sour grapes but Kasparov has reason to be suspicious. There's an old hacker saying that goes "any sufficiently advanced technology is indistinguishable from a rigged demo." Chess masters are not the first to get the feeling that their talents for merging of left- and right-brain thinking may be usurped by a few ounces of silicon and decades of research. Back in 1966, psychotherapists were given quite a scare by a program called ELIZA. ELIZA, created by MIT artificial intelligence researcher Joseph Weizenbaum, simulated a session of Rogerian psychoanalysis. The theory behind <A HREF="http://www.wynja.com/giganto/psych/rogerst.html">Rogerian psychology</A> is the therapist should remove as much of him/herself from the counseling session. The trick is to take a patient's previous statement and rephrasing it in the form of a question (I hate flying./What is it about flying that you hate?). It was a very enlightened, very `60s idea but totally useless. Studies show Rogerian psychoanalysis is about as effective as talking out your problems with a bartender. ELIZA is one of those classic programs, like Life and Star Trek, that get reimplemented every few years in the current "hot" language. There are various Java versions floating around on the web these days. For example, point your browser to the <A HREF="philly.cyberloft.com\bgoerlic\eliza.htm">ELIZA as a JAVA Applet</A> home page. If your browser doesn't support Java, you can get DOS and Mac versions at <A HREF="ftp://eecs.nwu.edu/pub/eliza/">ftp://eecs.nwu.edu/pub/eli za/</A>. Weizenbaum did not intend ELIZA to replace human psychotherapists. He was simply trying to show computers could process natural language. You don't need obscure commands like "dir". Weizenbaum did his best to distance himself from subsequent researchers who thought this was the next big thing in psychotherapy. His efforts, however, did not stop MIT denizens from firing up ELIZA and pouring out the most intimate detail of their life to a command-line prompt. People not part of the elite few with access to computers back in the '60s would actually call up Weizenbaum and beg him for time with ELIZA. You might snicker but people do something similar today over IRC. Is that 25-year-old woman you're chatting with really a woman or a 15 year old prankster? Is that sweet-talking 28- year-old guy really an advertising executive or an unemployed 48-year-old married man? Or is it really an updated version of ELIZA, a "bot", someone has turned loose on an IRC channel. It happens. IRC chatter bots are considered by some the ultimate hack. In 1950 mathematician <A HREF="http://www.wadham.ox.ac.uk/~ahodges/Turing.html">Alan Turing</A> (1912-1954) threw down a gauntlet in a paper titled "Computing Machinery and Intelligence". In it Turing proposed something that came to be known as the Turing Test. The paper argued if a computer could fool a human into thinking he or she was communicating with a human instead of a computer, the computer could be said to be intelligent. Now, if you're kind of going "huh?" at this moment, join the club. Most people have problems with Turing's rather bare-bones definition of intelligence. The Turing Test would have been dismissed as one of those quaint ideas -- much like the quaint idea that the world was only ever going to need five computers -- if it were not for the fact Turing was one of the 20th century's greatest geniuses. What Einstein was to physics, Turing was to math and computer science. Turing was key in cracking the German Enigma code, which most historians agree was one of the major tur(n)ing points in the war. Not many people outside of academia know much about Turing or his important role in winning the Big One. Turing probably would be a lot more famous if he had mustered the good sense to be born American and straight. Alas, history's hype machine has a way of ignoring British homosexuals who win wars by thinking too much instead of blowing stuff up real good. Computer scientists, however, have not forgotten Turing's raw genius and his many important contributions to the early days of their field. While the Turing Test may not be considered by anyone in AI research today as benchmark of machine intelligence, it has become something of an intellectual exercise -- much in the same way chess no longer teaches one much about modern warfare but teaches one a lot about logic, pattern recognition, and intuition. That the Turing Test is predicated on deception has also likely contributed to its popularity. Programmers are, as a rule, a mischievous lot.