NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/comp/ai/philosop/7182

Internet Message Format | 1992-12-31 | 4.1 KB

Path: sparky!uunet!zaphod.mps.ohio-state.edu!moe.ksu.ksu.edu!kuhub.cc.ukans.edu!spssig.spss.com!markrose
Newsgroups: comp.ai.philosophy
Subject: Re: grounding and the entity/environment boundary
Message-ID: <[e-mail]>
From: [e-mail] (Mark Rosenfelder)
Date: Thu, 31 Dec 1992 19:17:14 GMT
Sender: [e-mail] (Net News Admin)
References: <[e-mail]> <[e-mail]>
Organization: SPSS Inc.
Lines: 64

In article <[e-mail]> [e-mail] (Ken Easlon) writes:
>In article <[e-mail]> ,
>I plead guilty to bringing soup into the discussion, with no clear purpose
>other than to explore the ramifications of Chris Malcolm's idea that
>entity/environment lines can be drawn around "ownership".
>
>My involvement in this thread started when I read Mark Rosenfelder's
>comments to the effect that the entity/environment boundary should be drawn
>along contours of minimum information exchange.
>
>It seems to me there can be many shells within a functional definition of
>entity across which information exchange might be minimized, so I
>challenged Mark's criteria by bringing up human entities and tools, and
>suggesting an extreme example of a functional definition of entity where
>the boundary might be drawn along contours of maximum information exchange.
>
>This was the point where Chris suggested the ownership criteria and I
>brought in the soup.
>
>My current view as (ramblingly) posited in <[e-mail]>
>is that the entity should be defined as a core of tool operating
>capability, and interchangeable tools should be thought of as sort of a
>physical interface between the entity and environment.
>
>When Mark returns from vacation, he'll probably have some counter-arguments
>that will take the thread back to the issue of "grounding".

Only because I don't see how the entity/environment boundary can be discussed
in isolation; the subject came up in relation to grounding.  If I may quote
my own earlier salvo:

 I'm interested in what it takes to make an AI grounded-- to make its use of
 language meaningful rather than "merely syntactic"-- to have it know what
 it's talking about.  Placing Harnad, Lakoff, and my own musings into a
 blender and pressing "Liquefy", I come up with this: For an entity to be
 (statically) grounded it must have high-bandwidth sensorimotor experience
 with the real world; it's less grounded to the extent that the experience
 is of low quality or quantity, or not really its own, or is weakly
 integrated with its internal architecture.  Dynamic grounding would depend
 on a capacity to maintain this experience and function in the world.
 I trust you can see how this formulation leads to a concern with what is
 or isn't part of the system.

The problem with all the proposals aired so far, including my own, is a
maddening vagueness.  The ideal is would be criteria that can be applied
to the human case and produce the "right answer", which to me would be that
your eyes, fingers, and nose are part of you but your gloves, pocket knife,
and Unix account aren't; these criteria could then be applied to robots and
computers to help evaluate their grounding.

The kind of peripherals you've been talking about-- "interchangeable tools"
connected to the processor by a standard interface-- don't seem to me to offer
much grounding potential, precisely because they're low-bandwidth (grounding
requires loads of information) and because they're interchangeable (implying
low integration with the core entity).

Just in case I haven't laid out enough flame-bait already... Some alternative
conceptions of grounding avoid the boundary problem by concentrating on
correlation of the internal world model with the real world.  The problem I
have with these conceptions is that it's even less clear where you draw the
line between an ungrounded system that's just playing with symbols and one
that really knows what it's talking about; that it's not clear how many
errors, gaps, and ambiguities are permitted in the correlation; and that the
origin of the correlation is not taken into account.