Ed,
I have to side with Richard on this. The truth is *not* obvious. I
have my beliefs which I will express when I get the time later today or
tomorrow -- but -- there is absolutely no reason for you to be dismissive
like this. Richard is just *doing science* while you appear to me to be
*doing politics*. Your ideas *might* be better but his methods are MUCH
better.
Mark
----- Original Message -----
From: "Ed Porter" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Thursday, April 24, 2008 8:35 PM
Subject: **SPAM** RE: [agi] THE NEWEST REVELATIONS ABOUT RICHARD'S
COMPLEXITY THEORIES
I agree Richard, lets bury the hatchet on this one. I think the truth is
obvious to any one who has read your blog and followed the various thread of
this argument on this list. Ed Porter
-----Original Message-----
From: Richard Loosemore [mailto:[EMAIL PROTECTED]
Sent: Thursday, April 24, 2008 6:36 PM
To: [email protected]
Subject: Re: [agi] THE NEWEST REVELATIONS ABOUT RICHARD'S COMPLEXITY
THEORIES
Ed,
This is just garbled. And insulting, too: you try to imply that I have
changed my stance, when in fact I have said the same thing throughout,
but you keep finding new ways to get completely confused about what I
have said.
My point has always been that all those systems only work to a limited
extent, and one reason they only work to a limited extent is that they
try to pretend that cognition is not complex.
I am not going to explain anything else to you, because whatever I say,
you get confused and then launch even more accusations than before.
Sorry, but this is too silly. Bye.
Richard Loosemore
Ed Porter wrote:
Richard,
The most important point in your below post is your newly introduced
limitation that your four features of design doom don't necessarily
prevent
design of many AI systems --- but are --- you believe --- very likely to
cause design doom in very large AGI systems --- such as human-level
systems
--- that are extremely complex (in the old fashioned sense) ---
particularly, once they have done a tremendous amount of self
modification,
in the sense of automatic learning and adaptation.
It actually makes sense that as a system becomes vastly more complex ---
as
I believe any human level AGI's will have to be --- that complexity --- in
the sense of systems that are hard to properly control --- might become a
problem.
But since you now have admit the four features of design doom haven't
doomed
design in some many current AI --- which presumbably includes current
large
AI's like SOAR and LIDA --- there is nothing in your blog (at least as of
last night) or your response below to provide any indication of how large
a
system has to be --- or how much of each of the four factors are required
--- for design doom to occur. And there does not appear to be anything
other than a hunch --- on your part --- as to at what size design doom
becomes nearly or actually inevitable.
The Googleplex arguably has the four design features of doom and it has
run
for roughly a decade adapting its indexes to arguably more information
than
many full AGI systems may over their lifetimes, and it has remained
remarkably stable. It has memory. It has development in the sense of
automatically adapting its indexes. It has Identity at least in the sense
of identifying individual users and presumably types of users for use in
placing adds. Finally it has non-linearity in much of its decision
making,
such as in handling word forms, etc.
The Googleplex is less likely to have design doom problems than some AGIs
--- but it s not clear by how much, since your blog provide no math for
estimating where --- in a design space having as dimensions the four
features of design doom (and perhaps other parameters) --- design doom is
likely to kick in and to what degree.
So Richard it does not appear your theory of Richard-complexity --- with
its
the four features of design doom and its concepts of "untouchability" --
as
it relates to AGI has added anything solid to the AGI community's
understanding other than that when we build large automatically running
AGI's there might well be complexity problems that may present very real,
and possibly, extremely difficult problems.
I have said for years --- long before I ever heard of Richard-complexity
---
that the only really big problem I know of in making human level AI ---
(other than getting the massively parallel and highly interconnected
hardware and the software tools to program on it) --- is getting it all to
work together well automatically. It doesn't appear your theory adds
anything to that except a greater degree of doubt about whether we humans
are smart enough to solve that problem.
----------
Now I will indicate my comments on other parts of your response.
====RICHARD====>
Ed,
You have put words into my mouth: I have never tried to argue that a
narrow-AI system cannot work at all.
(Narrow AI is what you are referring to above: it must be narrow AI,
because there have not been any fully functioning *AGI* systems
delivered yet, and you refr to systems that have been built).
====ED=========>
I did not put words in your mouth. When I used AI I was using the term to
include both narrow AI and AGI, since both are AI systems (So its not a
strained interpretation I was intending)
There have been multiple systems such as SOAR and, I think, LIDA which
seem
to apply a common automatic learning and behaving architecture to many
different types of problems, and thus can be considers AGI's.
====RICHARD====>
The point of my argument is to claim that such narrow AI systems CANNOT
BE EXTENDED TO BECOME AGI SYSTEMS. The complex systems problem predicts
that when people allow those four factors listed above to operate in a
full AGI context, where the system is on its own for a lifetime, the
complexity effects will then dominate.
====ED=========>
Well, I wish you had said this in your blog. It does not appear to make
any
limitation that design doom depends on system being as large as huge AGI
system. In fact it says, speaking of the Four Features of Design Doom:
"These four characteristics are enough. Go take a look at a natural system
in physics, or an engineering system, and find one in which the components
of the system interact with memory, development, identity and
nonlinearity.
You will not find any that are understood."
This language implies design doom would occur on any system with these for
features. Thus, it implies just the opposite of what you call the "point
of
your argument" in your paragraph above.
====RICHARD====>
..
-----Original Message-----
From: Richard Loosemore [mailto:[EMAIL PROTECTED]
Sent: Wednesday, April 23, 2008 10:15 PM
To: [email protected]
Subject: Re: [agi] Adding to the extended essay on the complex systems
problem
Ed Porter wrote:
Richard,
In your blog you said:
"- Memory. Does the mechanism use stored information about what it was
doing fifteen minutes ago, when it is making a decision about what to do
now? An hour ago? A million years ago? Whatever: if it remembers,
then it has memory.
"- Development. Does the mechanism change its character in some way
over time? Does it adapt?
"- Identity. Do individuals of a certain type have their own unique
identities, so that the result of an interaction depends on more than
the type of the object, but also the particular individuals involved?
"- Nonlinearity. Are the functions describing the behavior deeply
nonlinear?
These four characteristics are enough. Go take a look at a natural
system in physics, or an engineering system, and find one in which the
components of the system interact with memory, development, identity and
nonlinearity. You will not find any that are understood.
".
"Notice, above all, that no engineer has ever tried to persuade one of
these artificial systems to conform to a pre-chosen overall behavior.."
I am quite sure there have been many AI system that have had all four of
these features and that have worked pretty much as planned and whose
behavior is reasonably well understood (although not totally understood,
as is nothing that is truly complex in the non-Richard sense), and whose
overall behavior has been as chosen by design (with a little
experimentation thrown in) . To be fair I can't remember any off the
top of my head, because I have read about many AI systems over the
years. But recording episodes is very common in many prior AI systems.
So is adaptation. Nonlinearity is almost universal, and Identity as you
define it would be pretty common.
So, please --- other people on this list help me out --- but I am quite
sure system have been built that prove the above quoted statement to be
false.
Ed,
You have put words into my mouth: I have never tried to argue that a
narrow-AI system cannot work at all.
(Narrow AI is what you are referring to above: it must be narrow AI,
because there have not been any fully functioning *AGI* systems
delivered yet, and you refr to systems that have been built).
The point of my argument is to claim that such narrow AI systems CANNOT
BE EXTENDED TO BECOME AGI SYSTEMS. The complex systems problem predicts
that when people allow those four factors listed above to operate in a
full AGI context, where the system is on its own for a lifetime, the
complexity effects will then dominate.
In effect, what I am claiming is that people have been masking the
complexity effects by mollycoddling their systems in various ways, and
by not allowing them to run for long periods of time, or in general
environments, or to ground their own symbols.
I would predict that when people do this "mollycoddling" of their AI
systems, the complex systems effects would not become apparent very soon.
Guess what? That exactly fits the observed history of AI. When people
try to make these AI systems operate in ways that brings out the
complexity, the systems fail.
Richard Loosemore
P.S. Please don't call it "Richard-complexity" .... it has nothing to
do with me: this is "complexity" the way that lots of people understand
the term. If you need to talk about the concept that is the opposite of
simple, it would be better to use "complicated". Personalizing it just
creates confusion.
-------------------------------------------
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/?&;
Powered by Listbox: http://www.listbox.com
-------------------------------------------
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/?&;
Powered by Listbox: http://www.listbox.com
-------------------------------------------
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/?&;
Powered by Listbox: http://www.listbox.com
-------------------------------------------
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/?&;
Powered by Listbox: http://www.listbox.com
-------------------------------------------
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=101455710-f059c4
Powered by Listbox: http://www.listbox.com
