home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!decwrl!decwrl!netcomsv!netcom.com!vere
- From: vere@netcom.com (Steven Vere)
- Subject: Re: Loebner Turing test for Human like conversation
- Message-ID: <1992Nov6.030414.20479@netcom.com>
- Keywords: Loebner, turing, natural language
- Organization: Netcom - Online Communication Services (408 241-9760 guest)
- References: <1992Nov5.212849.7986@cs.ucla.edu>
- Date: Fri, 6 Nov 1992 03:04:14 GMT
- Lines: 42
-
- In article <1992Nov5.212849.7986@cs.ucla.edu> colby@oahu.cs.ucla.edu (Kenneth Colby) writes:
- >In the summer issue of AI Magazine, there is an article on the Loebner
- >Prize competition for most human-like computer conversationalist, wriiten by
- >Robert Epstein, director Cambridge Center for Behaviorial Studies.
- >
- >The competition involves Turing-like tests in using ordinary natural language
- >in a specific domain. Daniel C. Dennett is chairman of the prize committee.
- >The next competition will be held 12/15/92.
- >
- >Does anyone have any comments, criticisms, or suggestions about the article
- >or the competition.
- >
- >Ken Colby - Malibu Artifactual Intelligence Works
-
- In my opinion the present Loebner contest is unrealistically
- difficult. Adult level conversation is just too sophisticated for
- the present state of the art, even when the subject is restricted.
- I have suggested that instead they try a preschool Turing test. In
- this variant, the human contestants are three year olds, and a parent
- reads them the question from the screen and types their response, if
- any. Three year olds have very simple minds, and can understand and
- use only very simple English. Their world knowledge is also extremely
- limited. Yet for all their limitations, they have functioning
- conscious minds.
- I think it is possible to develop a deep AI program in the near
- future that would perform decently on the preschool Turing test. Once
- the test is passed for three year olds, the age of the children could
- be progressively raised for future tests, until eventually the AI
- critics surrender. The full adult Turing test is a reasonable long
- term objective, but we need to crawl before we try out for the
- Olympics.
- --
- _____________________________________________________________________
- __ __ __ __ __ __ __ __ __ __ __
- |__| |__| |__| |__| |__| |__| |__| |__| |__| |__| |__| |
-
- Steven Vere Cold War Surplus Computer Scientist
- Boulder Creek, California vere@netcom.com
- * Artificial Intelligence Research and Seasoned Firewood *
- __ __ __ __ __ __ __ __ __ __ __
- |__| |__| |__| |__| |__| |__| |__| |__| |__| |__| |__| |
- _____________________________________________________________________
-