Newsgroups: comp.ai,comp.arch,comp.speech,rec.arts.sf.science,sci.cognitive
Path: cantaloupe.srv.cs.cmu.edu!nntp.club.cc.cmu.edu!miner.usbm.gov!news.er.usgs.gov!jobone!news2.acs.oakland.edu!newsxfer.itd.umich.edu!newsxfer3.itd.umich.edu!newsfeed.internetmci.com!Quza.UK.peer!uunet!in2.uu.net!204.74.114.90!news.genuity.net!news.lafn.org!lafn.org!lafn.org!ba672
From: ba672@lafn.org (Arthur T. Murray)
Subject: Re: modeling thinking (was Re: AI wants to grow up, but should it?)
X-Nntp-Posting-Host: lafn.org
Message-ID: <1997Jan29.151251.20000@lafn.org>
Sender: news@lafn.org
Reply-To: ba672@lafn.org (Arthur T. Murray)
Organization: The Los Angeles Free-Net
References: <5clme4$cku@pith.uoregon.edu> <Pine.SOL.3.95.970128180209.1384B-100000@butch.dai.ed.ac.uk> <5cllkn$u61@news.cei.net>
Date: Wed, 29 Jan 1997 15:12:51 GMT
Lines: 50
Xref: glinda.oz.cs.cmu.edu comp.ai:43839 comp.arch:74797 comp.speech:12172 sci.cognitive:14605


Plus royaliste que le roi, Lee Kent Hempfling <lkh@cei.net>:

>> ... to make a thinking machine one has
>> to actually think like the model not like the machine.

So we propose a Standard Model of thinking in man or machine.

  /^^^^^^^^^^^\ Syntax Strings Together a Thought /^^^^^^^^^^^\
 /visual memory\           ________   semantic   /  auditory   \
|      /--------|-------\ / syntax \  memory    |episodic memory|
|      |  recog-|nition | \________/<-----------|-------------\ |
|   ___|___     |       |     |flush-vector     |    _______  | |
|  /image  \    |     __|__  / \  _______       |   /stored \ | |
| / percept \   |    /     \/   \/ Verbs \------|->/ phonemes\| |
| \ engrams /<--|-->/ Nouns \    \_______/      |  \ of words/  |
|  \_______/    |   \_______/-------------------|-->\_______/   |

Voici le penseur M. L. Ginsberg <ginsberg@dt.cirl.uoregon.edu>:

> [...] The evidence is that to make a thinking machine, you have
> to think like the machine and *not* like the model.  Airplanes,
> after all, do not flap their wings.

  A gift for expression, and a sense that the world is not mad.

> Virtually all of the evidence from AI is that machines that think
> will do so by exploiting the fact that machines are fast.  [...]

  Non, non, monsieur l'ingenieur, dans le printemps de l'intelligence
  artificiel we will make a mind-modeled machine that thinks SLOWLY;
  apres le fait accompli we will make the AI mind de grande vitesse,
  the exploitation of both speed and massive parallelism:  Super AI.

> This probably says something fundamental about both men and
> machines. Our strength is pattern matching; theirs is search.
            ^^^ ^^^^^^^^ ^^ ^^^^^^^ ^^^^^^^^  ^^^^^^ ^^ ^^^^^^
  A too straitjacketed, too rigidly engineero-typical idee fixe.

> Perhaps this should not be so surprising after all -- pattern
> matching appears to be a fundamentally parallel process, while
> search is fundamentally serial.  We are parallel machines
> operating at millisecond time scales, while computers are serial
> machines operating at picosecond time scales.  We're both just
> doing what our hardware appears to be built for.

>             Matt Ginsberg
>             http://www.cirl.uoregon.edu/ginsberg
  --
  http://www.newciv.org/Mentifex/ run periodic Mentifex Web search.
