🚀 go-pugleaf

In article <3439451c.7302711@news.earthlink.net>
           savainl@earthlink.net "Louis Savain" writes:

> In article <875924895snz@longley.demon.co.uk>,
> David@longley.demon.co.uk (David Longley) wrote:
> 
> >In article <34352c7c.1516199@news.earthlink.net>
> >           savainl@earthlink.net "Louis Savain" writes:
> >> 
> >> .........  I think you have a somewhat pessimistic view of the power
> >> of inductive learning.  Sure, it is a trial and error processes that
> >> can easily give rise to superstitions, as you pointed out, but this
> >> trial and error process should not be underestimated.  It can also be
> >> used to send men to the moon and invent intelligent, conditionable
> >> robots.
> >> 
> >> Regards,
> >> 
> >> Louis Savain,
> >> 
> >You're  correct in your first sentence - and you are also  correct  in 
> >saying  that my view of "intelligence" is somewhat idiosyncratic.  But 
> >that doesn't mean it's wrong of course - maybe just unfamiliar. 
> >
> >I suspect we, as a species, have had ANNs for many thousands of  years 
> >- and that it was something such as the discovery of deduction  itself 
> >(and  especially the predicate calculus quite recently) which gave  us 
> >the  skills we regard as intelligent behaviour. Certainly I'd put  our 
> >ability  to  get  to  the moon etc.  down  to  deduction  rather  than 
> >induction.
> 
>   I agree, and as I wrote in a response to Neil Rickert, we could not
> have discovered deduction were it not for the power of our inductive
> learning capabilities.  In the context creating a general learning AI,
> I believe that induction is the only process that is fundamental to
> learning.  Our future AIs will discover deduction just like we did.

My point was that there seems little point in doing so as we know 
this  as we have already discovered deduction. Let me  illustrate 
my point with a rerasonably well known 'joke':

    'In the days when Sussman was a novice Minsky once  came 
    to  him as he sat hacking at the PDP-6.  
    "What are you doing?", asked Minsky.
    "I am training a randomly wired neural net to play Tic-Tac-Toe."
    "Why is the net wired randomly?", asked Minsky.
    "I do not want it to have any preconceptions of how to play"
    Minsky shut his eyes,
    "Why do you close your eyes?", Sussman asked his teacher.
    "So that the room will be empty."
    At that momment, Sussman was enlightened.'

> 
> >[...]
> >
> >My general conclusion is that *many* psychologists have in  gfact 
> >misread what their discipline has all been about since the 1960s. 
> >As an ampirical discipline, and in contrast to the technology  of 
> >AI,  it  is  a catalogue of the failings  of  a  system  (*human* 
> >information    processing)   optimised   for   fault    tolerance 
> >biologically,  but minimally rational without the support of  the 
> >extensional stance - artificial intelligence (science).
> 
>   I agree.  However, this extensional stance is learned inductively
> and we must acknowledge this fact if we are to create a general
> synthetic intelligence.  Inductive learning is a powerful way (I would
> argue that it is the only way) to quickly gain and manage large
> amounts of knowledge and, given enough time, this knowledge can become
> highly accurate in its agreement with nature.  We could not have
> survived otherwise.

I  agree that we learn through our experience, and that we  learn 
via  conditioning. But what we learn is the important point I  am 
drawing attention to, along with what we are able to do with what 
we  learn. The advances in science are based on observations  and 
what  one  does  with those observations  using  the  extensional 
stance.  We may  have ultimately learned this the hard way (as  a 
species) but a lot of what we have achieved is based on  applying 
the  extensional  stance to the data we have  collected  instead. 
There  is, in brief, a differenmce between (folk) psychology  and 
science, and it is the latter which is intelligent behaviour.

> 
> >[Well researched quotes deleted]
> > 
> >Throughout, it  is  being argued that such inductive inferences are in 
> >fact best modelled  using artificial  neural  network technology,  and 
> >that   such  processing  is intensional, with all of  the  traditional 
> >problems of intensionality:
> 
>   Agreed.  Our future AIs will need to be trained to follow rigorous
> empirically-grounded methods of research.  Most of them however will
> not be pressed into scientific service and their fallible "intensional
> processing" (the high level operation of which we will never fully
> understand) will work fine in the more mundane tasks for which they
> will be conditioned.  IMO, this is the real goal of AI.  I would agree
> with you that this goal cannot be achieved without us humans adopting,
> as an operating principle, the "extensional stance" that you are
> championing.
> 
>   Thanks for the pertinent quotes and the references.
> 
> Regards,
> 
> Louis Savain
> 
> President, Marengo Media, Inc.
> 
> 
You're welcome....good luck with the research.
-- 
David Longley (check end reply line #)

Longley Consulting                                                  London, UK
Behaviour Assessment & Profiling Technology,
Research, Data Analysis and Training Services,
Small IT Systems                                http://www.longley.demon.co.uk