Re: [tuning] Digest Number 2060

> From: "M. Edward (Ed) Borasky" <znmeb@aracnet.com>
> Subject: Re: Re: I'm baack :) for a while anyhow
>
> On Tue, 21 May 2002, Carl Lumma wrote:
<snip>
>
> > Yep, this kind of wrong thing. Surprising coming from Hofstadter, a
> > man who for any other aspect of thought, including Turing tests, has
> > no problem admitting that computers will do it, and has even done
> > important work in getting them to!
>
> Considering the kind of compute horsepower involved in, say, today's
> chess engines, I'm not at all convinced that *I* will live to see a
> computer pass the Turing test.
<snip>

Computers have already passed the Turing test. The problem
is that they keep moving the goal lines. There are already
programs that can fool non-experts into thinking they are
talking with a human. Certainly the least bright among
humans can be fooled by very simple programs. The problem is
expectations. Modern researchers are used to a level of
"humanity" in interactive programs that would have astounded
Alan Turing, and maybe fooled him for a couple of minutes.
Do you recall the "reverse Turing test" in one of
Hofstadter's articles, where a student at a terminal tried
to convince Doug that she was an advanced AI machine? He was
fooled for a few minutes, and I would guess it was because
of his expectations.

John Starrett