NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/comp/ai/philosop/7191

Text File | 1993-01-02 | 3.7 KB | 77 lines

Newsgroups: comp.ai.philosophy
Path: sparky!uunet!tcsi.com!iat.holonet.net!ken
From: [e-mail] (Ken Easlon)
Subject: Re: grounding and the entity/environment boundary
Message-ID: <[e-mail]>
Organization: HoloNet National Internet Access BBS: 510-704-1058/modem
References: <[e-mail]>
Date: Sat, 2 Jan 1993 17:10:04 GMT
Lines: 66


In article <[e-mail]> ,
[e-mail] (Mark Rosenfelder) writes:

>...For an entity to be (statically) grounded it must have high-bandwidth
>...sensorimotor experience with the real world;

I agree that such high-bandwidth sensorimotor experience (HBSME) might be
necessary for the development of biological systems, for example the retina
and other components of the visual processing system.  But I suspect nature
relies on HBSME because it's so readily and universally available as a
source of developmental information.  The evolution of a set of genes that
could shape the human visual system as accurately as HBSME might take a few
million years.

However, if we focus on the mature visual system, I think we could
duplicate the functions electronically, without using HBSME.

>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).

As a biological entity, I accept the idea that much of my development
during the first few years (after birth) depended heavily on HBSME.
However I think the bulk of intellectual development that gives me the
sense of knowing what I'm talking about arises from conceptual activity.

I think the mental source of new ideas (like dreams and inspirational
flashes) is a high bandwidth "entity" within the mind, call it high energy
imagination.  But I also think that much of the sense of knowing what I'm
talking about involves the mental manipulation of fixed concepts in a
relatively systematic way, call it rational thought.

In my view, rational thought is a low bandwidth operation, and can be
recreated on a computer with relative ease.  I would say the complexity or
bandwidth of rational thought is roughly the same as the bandwidth of
language usage.  Our internal perception of rational concepts is certainly
different than our perception of their word names, but the rules for
manipulating and developing rational ideas are very similar to the rules
for manipulating and developing verbal statements.

>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.

I think the proof is in the pudding.  If the AI does it's job, then it's
grounded enough.  Suppose the job of the AI is to develop a line of
rational thought, say participate in a debate.  I think we agree that the
AI can't perform this task just by using the names of concepts, but must
use some kind of program structure that is analogous to human thinking.  If
a human audience thinks the AI can hold its own in a debate with humans,
then I say it passes the "grounding" test.


--
Ken Easlon                | "...somebody spoke and I went into a dream..."
[e-mail]           |    -Paul McCartney
Pleasantly Unaffiliated   |