🚀 go-pugleaf

In article <1992Oct15.172441.30236@mp.cs.niu.edu> rickert@mp.cs.niu.edu (Neil Rickert) writes:
>In article <7707@skye.ed.ac.uk> jeff@aiai.ed.ac.uk (Jeff Dalton) writes:

>>The situation we're in right now with the TT is that of people
>>defending the visual test for gold in the absence of sufficiently 
>>good reasons for concluding that it works.  

>>Of course, there are differences between gold and intelligence
>>(or whatever).  I don't mean for the analogy to be exact.  (For
>>instance, we don't know yet whether "fools consciousness" is
>>actually possible.)

>Exactly, and that makes all the difference.  The visual test of gold
>would still be accepted as adequate if fools gold were merely a theoretical
>possibility which was not know to actually exist.

But machines that pass the TT are merely a theoretical possibility!
That's the sort of thing we're discussing here!

What I'm saying is that although we might eventually know enough
to know the TT works, we don't know enough now.  This seems pretty
straightforward to me.

Now you say "the visual test of gold would still be accepted".
This looks to me like an argument for the TT -- that we should
"still accept" it.  I realize that it doesn't _have_ to be an
argument for the TT, but it's certainly an atempt to defend the
TT from an attack rather than, say, a suggestion on how to make
the attack more effective.

>Perhaps usenet article get distorted when they cross the atlantic.  But
>I really have not seen many arguments - good or bad - for the TT.  I do
>see people pointing out the errors in frequent attacks on TT. 

Often trivial errors.  Why don't they say "your argument is almost
right and you could fix it thus"?  Of, for less trivial errors, "right
conclusion, bad argument; here's a better one"?

>And I see you misinterpreting these articles as if they were arguing
>for the TT.

I see people having plenty of opportunities to agree with other people
and not taking them.  If they agree with the conclusions these people
reach, why isn't there more of the sort of thing I just mentioned
above?

You know, it might be pretty easy to find concrete evidence that I'm
wrong.  There was recently a series of articles about the TT in
Comp.ai.philosophy (referring, for instance, to some recent stuff in
SIGART).  Someone could look at them and see what sorts of things were
said.  I read only a few of them, so I don't know how this will turn
out; and, as I said to Alan Smaill, I'd be glad to find I'm wrong.

>I think there are very few people in this discussion who would not agree
>that the internal dialogue is important.  However some of us believe that
>the internal dialogue does affect behavior, and therefore is tested with
>behavioural tests. 

Like what?  You can look at my behavior and tell if I'm thinking
about the weather?  Thinking about anything at all?  How?  Why
aren't more people making money as mindreaders?

-- jd