🚀 go-pugleaf

In article <7708@skye.ed.ac.uk> jeff@aiai.ed.ac.uk (Jeff Dalton) writes:

   In article <SMAILL.92Oct13181836@tummel.aisb.ed.ac.uk> smaill@aisb.ed.ac.uk (Alan Smaill) writes:
   >In article <7670@skye.ed.ac.uk> jeff@aiai.ed.ac.uk (Jeff Dalton) writes:
   >
   >   Why is it then that when I suggest that functional organization 
   >   (or, more generally, "how it works") should count, no one on the
   >   anti-Searle side says they agree?  The response has been so
   >   universally negative that I've more or less given up on trying
   >   to make that point.  

   >Maybe this has something to do with net.attitudes ? (the purpose of
   >posting is of course to point out where someone else got it wrong,
   >not where they got it right, after all).

   I think that's a factor.  But, for instance, supposing it's true that
   Daryl McCullough is a (the only?) behaviorist.  Why is it that so few
   people on the "pro-AI" side take issue with his arguments?

   Part of the reason may be another factor which we might call
   "temporary alliances of expediency" or "the enemy of my enemy
   is my friend".  But I don't think this would happen unless
   there was at least some agreement between the people on the
   same side.

And people do get defensive when they read certain arguments as
saying "your discipline doesn't make sense". 

By the way, I have trouble seeing the discussion here in terms
of "sides" at all.  On which side is a position like functionalism?
I see some variant of that as the commonest starting point for
AI people these days.

   Now, I'd be happy to be wrong.  If, in fact, most people on the
   net, or in AI, think the Turing Test is bogus, that would be great.

   But there is a fair amount of evidence pointing the other way.
   For instance, I said the other day that students in Edinburgh
   would argue that the Eliza program had a small amount of real
   understanding, and that the person teaching the course might
   argue on the same side.  Can they really be thinking of the
   functional organization of Eliza?  If so, why wasn't that clear
   in their arguments?  (This isn't true of everyone in Edinburgh,
   of course, and it was some years ago, but it's one of the reasons
   I started being more sympathetic to the arguments against AI.)

Students should of course be exposed to all sorts and conditions
of opinions, and learn early on that lecturers are not infallible...

   Or consider all the people who express the systems reply as "the
   system understands".  This is a minor detail, but what are people
   thinking when they say it?  Are they thinking `a la McDermott
   that the reason for saying the system understands would be the 
   theory used to construct it?  If so, why didn't they say so?

Well, a problem here is that while we can relatively  easily sketch
out the shape of a purely behaviour-based position, there is no
consensus on a position between that and Searle.  Indeed such a position
requires us to think very hard about the notions of understanding,
intelligence and so on, for which we have intuitive readings within a 
certain range of application, but which we are trying to use in situations
where our intuitions are less strong and maybe misleading.  

This is why I think that discussion of what we mean by these notions,
while it can easily degenerate into "define your terms" squabbles,
is important for the debate.

--
Alan Smaill,                       JANET: A.Smaill@uk.ac.ed             
Department of Artificial           ARPA:  A.Smaill%uk.ac.ed@nsfnet-relay.ac.uk
       Intelligence,               UUCP:  ...!uknet!ed.ac.uk!A.Smaill
Edinburgh University.