Newsgroups: comp.ai.philosophy,comp.ai,comp.robotics
From: ohgs@chatham.demon.co.uk (Oliver Sparrow)
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!EU.net!uknet!sunsite.doc.ic.ac.uk!pipex!demon!chatham.demon.co.uk!ohgs
Subject: Re: Minsky's new article
References: <39d8g2$dlm@coli-gate.coli.uni-sb.de> <39eaqk$nn9@cantaloupe.srv.cs.cmu.edu> <gyroCysG7u.8Hs@netcom.com>
Organization: Royal Institute of International Affairs
Reply-To: ohgs@chatham.demon.co.uk
X-Newsreader: Demon Internet Simple News v1.27
Lines: 80
Date: Tue, 8 Nov 1994 17:38:23 +0000
Message-ID: <784316303snz@chatham.demon.co.uk>
Sender: usenet@demon.co.uk
Xref: glinda.oz.cs.cmu.edu comp.ai.philosophy:21803 comp.ai:25054 comp.robotics:15142

Both camps are right. Let me come to this crabwise.

Knowledge comes in the declarative and tacit flavours. Declarative knowledge
can be explicitly captured, written down, tranmitted without loss. One can
read a book on how to change a transmission, program in C, fix a plug and, 
more or less falteringly, perform the task. Tacit knowledge cannot be so 
transmitted: one can offer hints on wine appreciation, on how to play a violin, 
on how to excite an audience; but in having read the best composed book 
conceivable, few who could not already do soi would feel that they can pick up 
their bow and begin to fiddle. 

This issue is a major problem for fields - such as commerce - in which 
*judgement* is more broadly appropriate than *analysis* as the basis of action. 
How is one to bring the distinct tacit models which people have across a team, 
a firm or an interest group into some modest harmony? Mostly, of course, people 
talk a lot to each other, but it is notable that in larger organisations, this 
will fail to serve. Innovation, for example, flows when those who have the 
experience and breadth of vision which comes from years of exposure the 
realities of affiars are able to encapsulate what they know in ways which set a 
challenge for those who - however narrow either circumstance or youth may have 
made the perception - nevertheless have acquired mastery of technologies, 
markets, networks and other  connectivity. Innovation arises when those who 
know what is possible - if only they knew what was wanted - come into dialogue 
with those who know what is wanted - if only they knew what was possible. At 
issue is how to solve this set of simultaneous equations on a machine made of 
people.

It turns out, therefore, that there are recognisable phases in which 
aspirational, visionary but focused statements of what might be are gradually 
clothed with the flesh of reality or consigned to the wind, as phantoms. It is 
necessary to have these visions, however, for without them, we are trapped with 
the pedestrian pursuit of ever-narrowing specialisation. Commercial research 
is, typically, governed by the disciplined and top-down parcelling out of 
chunks of work which contribute to and are assembled to make a broader whole. 
Academic science, by contrats, may often consist of the piecemeal exploration 
of what individuals deem to be interesting; and their results add up to 
coherence through a combination of social interaction and the fact that they 
are scraping coral off a common treasure: reality. Where this treasure does 
not yet exist - which is to say, in new product development, or, for example, 
in the creation de novo of artificial intelligence - then there **is** no 
buried treasure and the top-down schema must run. One must invent - at least 
in concept - the treasure for which one is hunting.

That Marvin feels it appropriate to strike a visionary note is, therefore, 
entirely valid. That others wish to seek substance - or a bridge which leads to 
a place where substance may be found - is also valid. What this Usegroup 
signally lacks, however, is any sense of an intermediary position. There are 
those areas of deep difficulty and of trembling, latent potential which one 
can, at least in principle, explore. 

There are accessible tools of exploration: one can envision what a saleable PC 
with AI built into it look like. Would it speak chirpily to me in the morning 
and think deep thoughts all of its own or would it, by contrast, have 
background systems which unobtrusively strove to understand what I had done and 
my motives then, what I was doing at present and my ostensible motives today 
and thus guided, set out to dig for relevant material? Would it be "sentient" 
or would it be useful? I have no desire for a system at home or in my car, 
traffic lights or in the collar of my dog which has attitude, needs to be 
wheedled, wants the vote... so I wonder if I *want* artifical sentience. What I 
probably want is an anticipatory heuristic which can search vast volumes of 
precedent, test for logical contradiction and remind me that I said that at 
*last* year's convention. So how do I get it? By contrast, I would deeply like 
to know how the brain works, how to think about the phenomenum of awareness and 
to understand ways in which this can be augmented; but this does not 
necessarily lead me to seeing an aware machine as the route to what I want from 
IT. It may be that there is a phase in the evolution of sentients when they 
cast aside biological things and find better seats of awareness in something 
which it would be hard to call a machine: a "matrix", perhaps. Doubtless inate 
in the nature of technological evolution, this is an interesting speculation 
but more than somewhat beyond the realms of the helpful guide for the issues of 
today.

Does anyone else want to try a bit of analysis of the nature of the prblem at 
which we are all hacking? What would the various facets of the various answers 
look like, if we were to envision them? Might this be a useful exercise?

_________________________________________________

  Oliver Sparrow
  ohgs@chatham.demon.co.uk
