From newshub.ccs.yorku.ca!torn!cs.utexas.edu!zaphod.mps.ohio-state.edu!darwin.sura.net!spool.mu.edu!umn.edu!The-Star.honeywell.com!saifr00.cfsat.honeywell.com!shanks Wed Sep 16 21:22:51 EDT 1992
Article 6867 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!torn!cs.utexas.edu!zaphod.mps.ohio-state.edu!darwin.sura.net!spool.mu.edu!umn.edu!The-Star.honeywell.com!saifr00.cfsat.honeywell.com!shanks
>From: shanks@saifr00.cfsat.honeywell.com (Mark Shanks)
Subject: Re: What is really AI?
Message-ID: <1992Sep10.201439.14233@saifr00.cfsat.honeywell.com>
Organization: Honeywell Air Transport Systems Division
References: <92254.103730KVJLC@ASUACAD.BITNET>
Date: Thu, 10 Sep 92 20:14:39 GMT
Lines: 40

In article <92254.103730KVJLC@ASUACAD.BITNET> Jon L. Campbell <KVJLC@ASUACAD.BITNET> writes:
>
>  For the sake of argument, I'd like to propose a discussion about
>  AI. Exploring what AI is or isn't and the boundaries of logic.
>
> Computational logic also has a set of inputs, yet solving mathematical
> problems does not use the same set of rules that were used to process
> common bodily functions (i.e. breathing)
> The set of rules required to breath are imbedded, basic rules
> that the brain always follows or we would forget to breath.  These
> rules are hardware dependent rather than learned and are therefore
> at the core of functional AI.  Learned rules are different in that
> they expand on the core layer of the brain or basic instructional set.

To make sure I understand: What you are saying is that hardware constructs
play an analagous role to physical (biological) constructs/survival
mechanisms with respect to AI, and that software/learning correlates to
human intellectual development. Is that close?

> How much software is needed to construct a truly intelligent unit?
> Is AI really software dependent?  Or can the hardware be independent
> of software to acheive AI?   Is software necessary?  Would hardware
> AI lack software as a counterpart?

Please forgive my editing: I want to address only some of the questions you
are asking, or maybe ask a few of my own. Is there NOW such a thing as
AI-dedicated software? (This is getting to be like a dog chasing his
tail; what is AI?) Let's assume (presume?) that Lisp, Prolog, and the
permutations of neural nets CAN be defined as AI-dedicated software in
that each is designed to manipulate SOME form of information in a manner
less "algorithmically" and more "relationship-oriented". This distiction
allows easier handling of "non-standard" inputs: the incoming data does
not necessarily need to be formatted in a specific way. Another benefit
is the ability of the software to alter itself (at least with NN) or
"learn". Hardware, on the other hand, can handle only what it is designed
to: the "inputs" are limited by the design. Hardware can't learn without
some software, and without learning, I don't think you can have AI. My
$0.02 worth.

Mark Shanks


