Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxl!ihnp4!zehntel!hplabs!sri-unix!Woody.pasa@XEROX.ARPA
From: Woody.pasa@XEROX.ARPA
Newsgroups: net.ai
Subject: Turing tests
Message-ID: <651@sri-arpa.UUCP>
Date: Mon, 30-Jul-84 13:25:00 EDT
Article-I.D.: sri-arpa.651
Posted: Mon Jul 30 13:25:00 1984
Date-Received: Sun, 5-Aug-84 00:24:05 EDT
Lines: 25

There's this accounting computer at the Santa Fe, (where my dad works),
and before it was installed, accounting was something which needed a very
intelligent person to do.  It required a high level of intelligence to keep
the books balanced, the type of intelligence a machine could never have.
The Santa Fe uses a computer to keep all their books now.
But note that the discussion with accounting is now not "The computer is
intelligent--look, it can keep the accounting books for an entire company",
but "Gee, anyone can keep the accounting books; even a computer."

The Turing test is a poor test, granted; but can there be a more
generalized test to tell if a computer is truly intelligent?  With the
Turing test, we can give the computer and the human at the other end a
test in math, understanding, and creativity; we could even talk about
the presidential elections; we're not restricted to the things that
have been discussed earlier.

As for hooking up a camera to the computer and using visual
identification as a test for intelligence: I know of a few blind
people who would be hard-pressed to past that test.  Sure, it takes a
lot to be able to see, but then most mice can see, and some humans
cannot; does that make the mice smarter than the humans?

  - Bill Woody
    1-60 Caltech
    Pasadena, CA 91126