X-Message-Number: 11710
From: Thomas Donaldson <>
Subject: To David Crevier, some answers
Date: Mon, 10 May 1999 00:35:45 +1000 (EST)

To Robert Crevier:

One quite critical issue which your little story slides over is that of
just how well we will ever be able to simulate a world. I am doubting that
our simulations will do very well; sure, I believe in using them for
various kinds of training, etc --- but the idea of living in a simulation,
given the weakness of such simulations compared to the real world, seems
quite ridiculous.

The SHRDLU example you give actually supports me. It's fine to have a 
computer which talks like it lives in the real world, but so far using
such simulations to make a device (part of which would be a computer) 
able to get around in the REAL WORLD have failed. I believe that by taking
the right approach we can even solve that problem, but so long as we
fail to solve that problem we have just what I said: a computer system
running a program with no consciousness. (And even working in a REAL
room full of blocks of various sizes is only a small start --- though
as I remember SHRDLU it didn't even get that far).

Please understand that I am not criticising SHRDLU itself. We learn how to
do the big problem by first working on small ones. But SHRDLU is not
a system which deals with reality as well as any 5 year-old. There are
tons more things such a system must be able to do; and most of all,
reality cannot be totally predictable and no one can really claim that
he or she understands it completely.

			Best and long long life,

				Thomas Donaldson

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=11710