Steve Richfield wrote:
Richard,
On 5/18/08, *Richard Loosemore* <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> wrote:
Steve Richfield wrote:
<large snip>
With luck we can wring things out at this level. With a little
less luck, a couple of weeks of attempted high-level design will
lead you to these same conclusions. With no luck at all, you
will dismiss the need for high-level design guiding the
functionality of low-level modules and continue on your present
bottom-up path, and quite probably spend much of your life
working on this apparently impossible and useless module.
The above text is gibberish.
Matt seemed able to deal with it.
That does not persuade me to re-evaluate my statement.
You make wild, sweeping and sometimes incoherent statements about
people and their work, based on complete ignorance of what those
people are actually doing.
Not 100% complete ignorance. Can we agree on 95%? I do believe that AGI
is possible, but my agreement with any of the viewpoints about how an
AGI might be constructed stops with that simple statement. I see lots of
counter-evidence of any understanding of the real world of natural
language (and that ~50% of real-world sentences will be mis-diagrammed
using good grammar rules), of the structure of real-world problems (that
they are usually a mental phenomena and not a real-world phenomena). I
have enough experience with software that WORKS in these problematical
domains to make some accurate statements about the domains themselves,
even though you might disagree on my software views.
Look, we both criticize the 'establishment' line on how to do AGI, but
if you are going to tell everyone they are wrong, you had better have a
cogent argument. If you have 95% ignorance about what is being done by
other people, don't criticize them.
Sure, cheap 'language understanding' systems are sometimes very bad.
But attacking them as if they represent the state of the art (not true)
or as if nobody is thinking about how to solve the problems (not true)
or as if there are no good techniques currently under development (not
true) is not constructive.
Your "Dr. Eliza" may be a modestly useful program, within its own
terms of reference. But it has nothing to do with AGI.
Allow me to restate the above in different terms that maybe we can both
agree on: There is nothing "intelligent" about Dr. Eliza - it is a
curious knowledge management program. It doesn't recognize patterns,
learn, experiment, argue, or do any of the usual things that we
attribute to "intelligence". However, it DOES accomplish some of the
more useful goals that people have been hoping that a future AGI can do
- some of the things that don't require recognizing patterns, learning,
experimentation, argument, etc.
My point is that there is no reason NOT to use what works now while we
work on something better, and the fact that Dr. Eliza does anything
useful at all tells us a LOT about the nature of complex problems, human
speech and writing, knowledge structure, what is currently missing on
the Internet and Wikipedia, etc.
My entire thesis was that the present AGI effort here really hasn't even
started the "analysis" phase, as evidenced by the lack of so many
considerations that were wrung out in creating Dr. Eliza. Nearly all of
the postings that I see here are about design issues, which cannot be
usefully approached until analysis has been substantially completed. You
appear to be discussing the details (that are in my 95% ignorance area)
to create modules that cannot work (for reasons that would be obvious in
analysis) and even if they could work, would have no place in an AGI
system (due to the incompatibility between correct- and
erroneous-paradigm statements).
My unsupported suspicion is that most of the posters here generally do
NOT understand foreign languages (some of which express the world VERY
differently), have NOT become world-class experts in multiple domains
(to see the extreme parallels between domains and problems therein), and
haven't even bothered to diagram a few dozen real-world sentences to see
that the grammar rules that everyone seems so willing to "bet the farm"
on simply don't work. In short, without a substantial breadth of
real-world knowledge and experience, there is simply no hope of
designing an AGI that isn't just as myopic. That my writings appear to
be gibberish to you is proof (at least to me) of this point.
It appears that you are on the "with no luck at all" path that I
outlined above.
Describe to me what my approach is. Give a summary of my paradigm, my
previous research, or the relationship between my approach to AGI and
someone else's. Give me, failing any of that, even a hint that you know
how long I have been working in this field.
You know absolutely nothing about what I am doing. Is that correct?
And yet you criticize it?
Some of us have known full well that 'grammar rules' do not describe the
actual speech (and colloquial writing) of people who communicate
meanings perfectly well. This is a standard piece of information that
anyone who studies psychologuistics knows from their first course on the
subject. Your tirades on the subject have about the same significance
as a solitary inventor bursting into a high-school physics class
shouting "Aristotle was wrong! Things keep moving in a straight line
unless a force acts on them!!!".
Richard Loosemore
-------------------------------------------
agi
Archives: http://www.listbox.com/member/archive/303/=now
RSS Feed: http://www.listbox.com/member/archive/rss/303/
Modify Your Subscription:
http://www.listbox.com/member/?member_id=8660244&id_secret=103754539-40ed26
Powered by Listbox: http://www.listbox.com
