Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!seismo!mcvax!ukc!its63b!hwcs!aimmi!gilbert From: gilbert@aimmi.UUCP (Gilbert Cockton) Newsgroups: net.ai Subject: Re: Searle, AI, NLP, understanding, ducks Message-ID: <813@aimmi.UUCP> Date: Tue, 21-Oct-86 09:29:09 EDT Article-I.D.: aimmi.813 Posted: Tue Oct 21 09:29:09 1986 Date-Received: Thu, 23-Oct-86 22:53:25 EDT References: <1918@well.UUCP> <1919@well.UUCP> Reply-To: gilbert@aimmi.UUCP (Gilbert Cockton) Distribution: net Organization: Heriot-Watt/Strathclyde Alvey MMI Unit, Scotland Lines: 79 In article <1919@well.UUCP> jjacobs@well.UUCP (Jeffrey Jacobs) writes: > >Most so-called "understanding" is the result of training and >education. We are taught "procedures" to follow to >arrive at a desired result/conclusion. Education is primarily a >matter of teaching "procedures", whether it be mathematics, chemistry >or creative writing. The *better* understood the field, the more "formal" >the procedures. Mathematics is very well understood, and >consists almost entirely of "formal procedures". This is contentious and smacks of modelling all learning procedures in terms of a single subject, i.e. mathematics. I can't think of a more horrible subject to model human understanding on, given the inhumanity of most mathematics! Someone with as little as a week of curriculum studies could flatten this assertion instantly. NO respectable curriculum theory holds that there is a single form of knowledge to which all bodies of human experience conform with decreasing measures of formal success. In the UK, it is official curriculum policy to initiate children into several `forms' of knowledge (mathematics, physical science, technology, humanities, aesthetics, religion and the other one). The degree to which "understanding" is accepted as procedural rote learning varies from discipline to discipline. Your unsupported equivalence between understanding and formality ("The *better* understood the field, the more "formal" the procedures") would not last long in the hands of social and religious studies, history, literature, craft/design and technology or art teachers. Despite advances in LISP and connection machines, no-one has yet formally modelled any of these areas to the satisfaction of their skilled practitioners. I find it strange that AI workers who would struggle to write a history/literature/design essay to the satisfaction of a recognised authority are naive enough to believe that they could program a machine to write one. Many educational psychologists and experienced teachers would completely reject your assertions on the ground that unpersonalised cookbook-style passively-internalised formalisms, far from being a sign of understanding, actually constitute the exact opposite of understanding. For me, the term `understanding' cannot be applied to anything that someone has learnt until they can act on this knowledge within the REAL world (no text book problems or ineffective design rituals), justify their action in terms of this knowledge and finally demonstrate integration of the new knowledge with their existing views of the world (put it in their own words). Finally, your passive view of understanding cannot explain creative thought. Granted, you say `Most so-called "understanding"', but I would challenge any view that creative thought is exceptional - the mark of great and noble scientists who cannot yet be modelled by LISP programs. On the contrary, much of our daily lives has to be highly creative because our poor understanding of the world forces us to creatively fill in the gaps left by our inadequate formal education. Show me one engineer who has ever designed something from start to finish 100% according to the book. Even where design codes exist, as in bridge-building, much is left to the imagination. No formal prescription of behaviour will ever fully constrain the way a human will act. In situations where it is meant to, such as the military, folk spend a lot of time pretending either to have done exactly what they were told or to have said exactly what they wanted to be done. Nearer to home, find me one computer programmer who's understanding is based 100% on formal procedures. Even the most formal programmers will be lucky to be in program-proving mode more than 60% of the time. So I take it that they don't `understand' what they're doing the other 40% of the time? Maybe, but if this is the case, then all we've revealed are differences in our dictionaries. Who gave you the formal procedure for ascribing meaning to the word "understanding"? >This leads to the obvious conclusion that humans do not >*understand* natural language very well. >The lack of understanding of natural languages is also empirically >demonstrable. Confusion about the meaning >of a person's words, intentions etc can be seen in every interaction ... over the net! Words MEAN something, and what they do mean is relative to the speakers and the situation. The lack of formal procedures has NOTHING to do with breakdowns in inter-subjective understanding. It is wholly due to inabilities to view and describe the world in terms other than one's own. -- Gilbert Cockton, Scottish HCI Centre, Ben Line Building, Edinburgh, EH1 1TN JANET: gilbert@uk.ac.hw.aimmi ARPA: gilbert%aimmi.hw.ac.uk@cs.ucl.ac.uk UUCP: ..!{backbone}!aimmi.hw.ac.uk!gilbert