X-Message-Number: 5776
From:  (Thomas Donaldson)
Subject: Re: CryoNet #5472 - #5474
Date: Tue, 20 Feb 1996 11:32:12 -0800 (PST)

Hi again!

The key issue about making "intelligent machines" is working out just what is
meant by intelligent. A machine is not intelligent just because it can do
something which I can't, even if that is something we ordinarily think of as
"mental". I can't even do much arithmetic without pen and paper, or a 
calculator. The intelligence lies in the creator of the computer or the 
calculator, not in the machines themselves.

If we start breaking down "intelligence" into specific abilities, it no 
longer looks so clear that consciousness is needed to do them. And maybe even
better than we can. But of course we don't consider the ability to do 
lots of arithmetic very fast to be "intelligence". Sometimes a component of 
"intelligence", but not the same. Other such abilities raise the same 
issue. Even the recent chess match between Kasparov and Deep Blue wasn't
really a match between Kasparov and a machine: it was a match between Kasparov
and the engineers who built Deep Blue.

One fundamental difference between Deep Blue and any living vertebrate is
very simple and maybe even profound. The animal is set up with its own desires,
and uses its brain as one means to achieve those desires. And it is the desires
and feelings which play the lead role here (in advanced animals, there is 
feedback, but desires and feelings still play the lead role). To put it 
very succinctly (perhaps too much) it is our desires and feelings which make
us intelligent, not our ability to remember, or calculate, or whatever: a
neat little paradox.

But I don't think that sums up "intelligence" even then. The word is put
about so much, but no one seems to know what it really means --- unless we
restrict the field (say to "intelligence tests") so much that in terms of
behavior we have very little. (No doubt another set of engineers, if they 
wished to, and if others agreed with these desires enough to provide the
funding, could make a machine to pass "intelligence tests". If you believe
that such a machine, just because it can do that, is "intelligent", you 
simply haven't thought out what you mean).

			Best to all, and a long long life,

				Thomas Donaldson


Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=5776