[Top] [All Lists]

Re: [ontolog-forum] Monica Anderson on AGI

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Sun, 03 Apr 2011 16:21:41 -0400
Message-id: <4D98D6D5.8040002@xxxxxxxxxxx>
Ali,    (01)

Thanks for the pointers.    (02)

> One criticism of this piece is that it only allocates a purist
> description of each approach (i.e. Cyc being a purist logical approach).
> Whereas efforts like IBM's Watson have demonstrated that multiple
> approaches are complementary and not exclusive (a point also argued
> persuasively by John Sowa in www.jfsowa.com/pubs/paradigm.pdf).    (03)

And thanks also for the citation.    (04)

But I'd like to add that logic is much more general than deduction.
Peirce, Whitehead, and every sensible logician since Aristotle
emphasized one very obvious point:  deduction starts from premises.
You can't depend on the truth of the conclusion if your premises
are faulty or your observations are mistaken.  You need some
source of new premises to try, which Peirce called 'abduction'
and some method for testing those premises against reality.    (05)

Monica's criticisms of purely deductive AI were made by C. S. Peirce
as long ago as 1887 in an article on "Logical Machines" in vol. 1
of the _American Journal of Psychology_:    (06)

http://projecteuclid.org/DPubS/Repository/1.0/Disseminate?view=body&id=pdf_1&handle=euclid.rml/1204900343    (07)

In those days, the only logical machines he could describe were
Babbage's design, which had never been implemented, a couple of
mechanical reasoning systems by Jevons and Marquand, and a parody
of mechanical contrivances by Jonathan Swift in _Gulliver's Travels_.    (08)

In that article, Peirce discussed the simple inference machines of his
day and the possibility of developing much more complex machines:    (09)

> And even if we did succeed in doing so,
> it would still remain true that the machine would be utterly
> devoid of original initiative, and would only do the special kind
> of thing it had been calculated to do. This, however, is no defect
> in a machine; we do not want it to do its own business, but ours.    (010)

Then he added:    (011)

> We no more want an original machine, than a house builder would
> want an original journeyman, or an American board of college
> trustees would hire an original professor.    (012)

This was a thinly disguised comment on his difficulty in finding
a teaching job.    (013)

> In part 2 of this series, [Monica] takes aim at the "Reductionist"
> stance, and presents four flavours of "impossible-to-Model" systems.
> She argues for a definition of intelligence based on: “The ability
> to perform Epistemic Reduction” and briefly hints at her own approach
> (Artificial Intuition), but the piece is mainly a critique of
> Reductionism and doesn't explicitly present any concrete constructive
> alternatives.    (014)

I would agree with her criticisms, and so would Peirce. But CSP
presented some very concrete alternatives a century ago.  And we
have been using his guidelines at our VivoMind company.    (015)

For a summary of Peirce's later views (around 1906), I drew a diagram
that illustrates his "cycle of pragmatism," which includes induction
from observations, abduction to form a new theory, belief revision
to revise and correct theories, deduction from the theory, and
action to test the conclusion against observation.  Then repeat...    (016)

See slide 37 of the following presentation:    (017)

    http://www.jfsowa.com/talks/situ.pdf    (018)

This would be Peirce's version of how intuition (natural or artificial)
works in "any scientific intelligence" -- which he did not limit to
humans.  He discussed dogs and parrots as examples, and he would not
exclude alien life forms, if they existed.  (Since he had spent several
years working at the Harvard observatory, he may have considered them.)    (019)

John    (020)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (021)

<Prev in Thread] Current Thread [Next in Thread>