Re: IP: Birth of a Thinking Machine

From: hal@finney.org
Date: Thu Jun 21 2001 - 11:06:03 MDT


I've been a little skeptical about CYC since reading this independent
review, http://boole.stanford.edu/pub/cyc.report, which did not find the
performance of the system to meet expectations. However that was back
in 1994 and hopefully they will have made a lot of progress since then.
Still the impression I had was that Lenat had somewhat oversold the
system's capabilities. I am afraid that the upcoming public release of
OpenCYC will be a disappointment.

It's kind of ironic, as chatbot technology has progressed considerably.
These systems have no real understanding but use conversational tricks
(changing the subject, echoing, distraction, flirtation, etc.) to fool
people into thinking there's someone there. Then here CYC comes along
and actually tries to do it right, but it may end up with some kind of
lumbering interface and not be as impressive as a 50K byte Basic program.

Hal



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:08:14 MST