From: Damien Broderick (d.broderick@english.unimelb.edu.au)
Date: Mon May 13 2002 - 21:22:57 MDT
At 03:11 PM 5/13/02 -0700, Ken wrote:
>The Turing test
>is one sided. It originated because no satisfactory definition of thinking
>could be established. It was generally recognized that humans think, so if
>something could pass the Turing test, then that thing should be recognized as
>being able to think as well. However, failure to pass the Turing test
does not
>establish that something cannot think.
Moreover, the Turing test appears to assume that in order to pass as a
human (under its constraining communication protocol), an AI would need to
emulate human internal processes. But it seems likely to me that *success*
in passing the Turing test--that is, convincing the bona fide humans that
they are talking with a human-capable intelligence--does not necessarily
establish anything about *how* that something thinks.
Damien Broderick
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:14:03 MST