home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai
- Path: sparky!uunet!think.com!ames!elroy.jpl.nasa.gov!ucla-cs!oahu.cs.ucla.edu!colby
- From: colby@oahu.cs.ucla.edu (Kenneth Colby)
- Subject: Re: Loebner Turing test for Human like conversation
- Message-ID: <1992Nov10.205230.28776@cs.ucla.edu>
- Keywords: Loebner, turing, natural language
- Sender: usenet@cs.ucla.edu (Mr Usenet)
- Nntp-Posting-Host: oahu.cs.ucla.edu
- Organization: UCLA, Computer Science Department
- References: <1992Nov5.212849.7986@cs.ucla.edu>
- Date: Tue, 10 Nov 92 20:52:30 GMT
- Lines: 20
-
-
-
-
- Robert's from Australia has it right about formalisms. Growing up in,
- and now saddled by, the Euclidean tradition that all problems can be
- solved with 5 axioms (the fifth turned about to be incorrect), many
- people in AI try to force emperical problems into a formal framework
- that doesnt get anywhere.
-
- Little progress has been made in AI with the "natural language problem"
- because of reliance on Aristotelian-Chomskyian categories, English
- grammers that are really Latin grammers, predicate calculus, theorem
- proving and other deductive dead ends.
-
- Language is more like a biological organism - "a summing up of many
- contrivances, each useful to the possessor" - C. Darwin.
-
- Ken and Pete Colby
- M.A.I.W.
-
-