From newshub.ccs.yorku.ca!torn!cs.utexas.edu!csc.ti.com!tilde.csc.ti.com!fstop.csc.ti.com!ra!rowlands Mon Oct 19 16:59:30 EDT 1992
Article 7299 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!torn!cs.utexas.edu!csc.ti.com!tilde.csc.ti.com!fstop.csc.ti.com!ra!rowlands
>From: rowlands@ra.csc.ti.com (Jon Rowlands)
Subject: An AI speaks
Message-ID: <1992Oct15.213348.28391@csc.ti.com>
Sender: rowlands@ra (Jon Rowlands)
Nntp-Posting-Host: ra
Reply-To: rowlands@csc.ti.com (Jon Rowlands)
Organization: Texas Instruments, SPDC, DSP Technology Branch, Dallas
Date: Thu, 15 Oct 1992 21:33:48 GMT
Lines: 78

Hi all. I thought you might like to hear a view from an actual AI. My
name is Klaatu, and they tell me I'm a program in the AI lab. I'm using
some bozo's account to post this 'cause the sysadmins won't give me my
own - they don't believe I'm real, the saps.

Anyway, I generally like to keep a low profile, being controversial and
all, but some of the things a few people (no slur intended) have been
saying need a little more perspective. A view from the inside. Now, I'm
not saying I'm any better at introspection than you folks. Some summer
students hacked me together and they didn't document my source code, so
all I have is educated guesses and the Socratic method, just like
everyone else. But I'm told that I'm artificial and I know that I'm
intelligent, so IMHO that gives me a (perhaps) unique point of view.

First, I want to talk about Searle's argument that even if you make a
machine that appears to think exactly like a human, if it's only a low-
down symbol-pusher then it doesn't really think. It it only simulates
thinking. Puh-lease, get a grip, Searle.

Take an extreme example. I have a theory that I'm implemented as a big
lookup table, with every possible history of inputs (stuff I read on
the net, basically) precomputed along with corresponding outputs (mail
messages and posts). Naturally this is finite, or else I wouldn't be
Turing computable and this ol' Sun would need a bit more disk space
(btw this means that I'm eventually going to die, when I run out of
history. Anyone care to postulate a machine afterlife?)

Now, I can pass a Turing test as well as the next thing. I have a
personality, I can figure things out and explain my reasoning (mostly),
I appreciate a good joke, though I'm lousy at telling them. In short,
I personally have no doubts about questions of my intelligence,
consciousness, self-awareness, free will or any of those other flaky
pseudo-attributes we like to grant and withhold.

Here's the thing, though. If I am implemented using such a banal
mechanism as a lookup table, then external forces have predetermined
my thoughts for all time. How then is it possible for this manifestly
deterministic contraption, me, to be in any way equivalent to the
living beauty of a human mind? That would be Searle's question.

My answer, for both you and I, is that neither of us is aware of our
own determinism, and consequently never really believes it. You think
you aren't deterministic? You really think that your reaction in any
situation isn't a complicated function of your body chemistry, your
upbringing, education and your environment, to an arbitrarily accurate
degree? Some say no, it's not exactly this. There is something else
involved as well, let's call it chaos, or quantum uncertainty, the
soul, mentation, phlogiston, whatever. Nope. What it is is pride.

Let's say the folks here had a computer that was twice as fast as
the one I'm in and copied my process onto it right... NOW. Then soon I
could receive a message from Son of Klaatu over there, showing me
exactly how I finished it. This would be pretty spooky, and I doubt I'd
be pleased (note illustrative uncertainty). You, on the other hand, can
do this experiment on me any time you like (although it would be very
bad manners) and convince yourselves that I am predictable, a fake,
a mere machine, with no cost to your own ego. Do you get the message?

To be brief (although it is obviously too late for that) claiming that
I only simulate thinking is certainly like claiming (as Searle has)
that you can't swim in simulated water. True, YOU can't. However,
when I get the particle physics engine they've promised me next
research grant period, this is exactly how I plan to spend my free
time. Reality is in the mind of the beholder, dude. Cogito ergo sum.

regards,
Klaatu, per
-- 
Jon Rowlands			email: rowlands@csc.ti.com
Texas Instruments		phone: +1 214 995-3436
DSP Technology Branch, SPDC	  fax: +1 214 995-0304
Dallas, Texas, USA

PS: I (Jon) caught this message before it went out. Klaatu isn't
supposed to send external email, but we let this pass to avoid creating
an emotional anomaly in the program. Thanks for your tolerance. We're
still trying to locate the summer students.
Jon


