Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!EU.net!sun4nl!hermes.bouw.tno.nl!usenet
From: sst@bouw.tno.nl (Tako Schotanus)
Subject: Re: The turing test
Message-ID: <1995Mar22.131806.24821@hermes.bouw.tno.nl>
Sender: usenet@hermes.bouw.tno.nl (USEnet Postmaster id)
Nntp-Posting-Host: ruudnix
Organization: TNO Bouw
X-Newsreader: WinVN 0.92.6+
References: <3kl0r8$id5@newssvr.cacd.rockwell.com> <3kloip$q9u@newsbf02.news.aol.com> <3kna8t$2tq@newssvr.cacd.rockwell.com>
Date: Wed, 22 Mar 1995 13:18:06 GMT
Lines: 183

In article <3kna8t$2tq@newssvr.cacd.rockwell.com>, csmccue@cacd.rockwell.com (Craig S. Mc Cue) says:
>
>In article <3kloip$q9u@newsbf02.news.aol.com>, mickwest@aol.com says...
>
>[quoted text reformatted somewhat]

>>
>>If computers could simulate sentience to such a degree that you could not
>>distinguish a computer from a human. Then how can you say that the human is
>>sentient and the computer is not.
>
>IF, that is...let me know when this vaporware is finally released. Your own 
>statments about subjective vs objective reality should suggest to you why we 
>shouldn't assume a computer is sentient just because we PERCEIVE it to be 
>sentient. Consider animism, folk magic and superstition. All these involve the 
>personification of inanimate things or results. Consider a gambler, who 
>considers luck a quantifiable substance that can be bargained with. There are 
>currently several hundred million Animists in the world today that believe that 
>trees and rocks have spirits, are sentient, and will get angry with you if you 
>disturb them. Does this mean that trees and rocks should be considered 
>sentient?

Well, we'd first need a realy good definition of sentient would we? And maybe
we can then tell if there would be a remote possibility of trees and rock being
sentient, if not we'd be in a position to convince all those people they're wrong.
But most likely it's improvable and we just take for granted that most people
in the western world think trees and rocks aren't sentient.
For example: I don't know if your religious but for the sake of argument let's
say you are. You'd tell me about your god and about christ and for you they are
real and important. I wouldn't think of disagreeing with you because I can't
prove to you that god *doesn't* exist even though that's what I believe.
So does this mean we should only believe in what is provable? I think most
people should stop believing most of what they believe because I think that's
what believing is most of the time: thinking to know something that can't be known.


>If I do not know what is going on in my mind when I am sentient, then how do I 
>know I AM sentient? Again, this goes back to a commonality of understanding 
>regarding sentience. 

So what you say is that you DO know what going on iside your mind?? Wow! Great!
You must be the first person on earth to be able to do so! Why haven't they made
you the subject yet of deep-probing investigations? We could learn so much from
you!

Of course this is light sarcasm: I can tell you now that you don't really know
what is going on inside your mind and saying you know you're sentient is just
giving a name to some vague feeling you couldn't even begin to describe fully.

>>> As stated above, yes we can dissect their innerds, as it were, and observe 
>>> how they think (ever hear of a debugger?)
>>
>>If a self learning program were to grow to such a size and complexity that it
>>could simualte human sentience, then to you really think you could understand
>>how it worked by looking at it with a debugger. I feel that when computers
>>programs get that complex, then only the computers will really know what is
>>going on.
>
>Again, this is vaporware...

That doesn't mean you can't discuss it :) We'd just have to agree or disagree
about the possibilty of such a computer ever being built or not.

>>
>>What I meant by "I don't know what is going on inside the computer" is that
>>it is infeasible for the computer to give me a state dump every time it makes
>>a response. Sure, It is possible, but in general, when talking to a computer,
>>you will not know what is going on inside it. 
>
>Having written a number of rather sophisticated simulators, modelling very 
>complex dynamic systems, I can suggest that such a monitoring of conditions is 
>possible, though difficult.

Than there are two possibilities:
- or the underlying model of this system is relatively simple and you're able to
  recognize similarities between the debugger output and the expected behaviour
  with regard to the model.
- or you have no idea whatsoever why your system is behaving as it is (compare:
  the human brain) it just happens to be more or less what you want/expect. If
  so then you *won't* be able to "digest" the debugger's output either because
  that would mean *understanding* what it's doing.

>>We are getting into meaningless realms here. What you are talking about are
>>abstract concepts. What is love? I'm sure you could get a computer to
>>simulate what you belive to be love and all these other instincts. But is it
>>'real' love? What is 'real' love? Perhaps if it looks like love, feels like
>>love and tastes like love, then it is love. But then perhaps it is not, you
>>really have no way of knowing as any definition of love is usually subjective
>>and meaningless.
>
>But computers, if they are to simulate or model sentience to us, must be able 
>to relate to our abstract concepts. I'm getting a bit confused here -- in one 
>point you suggest that subjectivity is the only reality we can relate to, but 
>here you say that it is meaningless. Please define your terms for me.

Meaningless in a way that love is never "understandable" and therfore only
debatable by beings experiencing it. Your example of the blind man: we can talk
about colors with him and he can ask questions about it but he would *never* be
able to make a meaningful statement about colors from his experience. So computers
might one day be able to discuss the *concept* of love but they would never
understand it. But nor do we, we just experience it. :)

But hey, maybe we'll one day understand how love "works" and be able to simulate it.
It's probably just a mixture of hormones anyway ;)

The thing with love being subjective and meaningless is probably that love is
just as difficult to define as sentient and that we're talking about wether
computers could ever be intelligent and if that would include sentience as well.
Love is another matter altogether and might be discussed if somebody wants to
make a computer that can love.

> And the fact that he got caught later with another prostitute 
>suggests that he did not have true repentance.

Or just that the "flesh is weak".

>Yeah, what you said... and, well, I CAN get pretty accurate in describing the 
>motion of the sun -- but ultimately it IS only a model, like the geocentric 
>and heleocentric models of the solar system. Point is, we relate to reality 
>through models, which are completely subjective from our viewpoint, but which 
>can also be demonstrated as having a very tight correlation to observed 
>reality.

No, the point is that you can't say what reality IS, most of the time it is just
what it APPEARS to be.

>
>>The reason I asked for an example was to show that even if you see through
>>what something seems to be, you have not seen what it "really" is. You have
>>simply move one level deeper in a series of vauge levels that eventually ends
>>with "Things are the precise position, mass, spin and velocity of every
>>partical in the universe" 
>>
>>You have to stop somewhere, because the ultimate level is ultimatly
>>unknowable. You usually stop at a useful model that works. But it is not
>>reality, just a model, just how things seem to be. 
>
>My point exactly. But you seem to insist that, because it is a model of 
>reality, it is no good, and thus we can never know anything, except through a 
>framework of perception. This seems to resemble the argument of the beard... 
>
No it's the other way around, you seem to insist that a model is more real if
it "explains" reality more thoroughly. The thing is that the other, "higher",
models might be just as "real" and therefore anything that can be described
by any of a number of models of a certain concept can be said to *be* that concept.

To go back to the earth/sun example: the model of the sun around the earth is
not less "true" than the model of the earth around the sun which again isn't
less "true" than the earth and the sun revolving around a common point.

which brings us back to computers: if we make a good model of intelligence and
sentience and we make a computer that implements those models it can be said
to be intelligent as well as sentient. But until the time that we have such
models we'll have to do with just the definitions of those concepts. But wait!
We don't even have good definitions yet, so how the h*** are we going to
evaluate proposals as how to make a computer intelligent/sentient?

This is something proposed by Penrose in his sequel to "The Emperors New Mind"
(forgot the title): he said that the first thing we should do is agree on what
in fact IS "intelligence", "self-awareness", "consience" and "sentience"
(or at least what it APPEARS to be *grin*)

>>My argument is not "Things are what they appear to be", since obviously they
>>are not. Rather the argument is "There is no possibility of knowing true
>>objective reality, so we can assume things are as they apppear to be, unless
>>evidence points to the contrary".
>>
>>Nothing is as it appears to be, EVERYTHING is simply "models usueful to us". 
>
>So then we get into dogmatic arguments about whose model is best? That seems to 
>be what is happening here... 

No I don't think that's what happening here. What he says is that when two
things can be explained by the same model they must be the similar and that
questions about reality or simulations are moot because there's no difference
between them.


_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
_/ Tako Schotanus                TNO Building and Construction Research _/
_/ Phone : +31 15 842393 Fax : +31 15 122182  E-mail : sst@bouw.tno.nl  _/
_/ My employer is required,by Dutch law,to disagree with whatever I say _/
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
