David Butler wrote:
I would say that the best way to simulate human intelligence with
diversity and creativity is to create not one AGI but many. The only way
to insure diversity and natural selection like our own evolution is to
simultaneously create multiple AGI's so that we have a better chance of
the emergence of the best path for the evolution of friendly AGI.
I am new to this list. Is there anyone out there who has addressed this
issue? We have many people who are very gifted with math and science who
are in the forefront of AGI, but "random" creativity and seat of the
pants intuition is a really big part of human evolution. If we create
multiple AGI's we have a chance that all of our traits are developed (in
the same way that we are genetically programed) in some way to create a
community of sorts that hopefully will be able to sustain our legacy of
diversity and creative thought.
Dave Butler
Making one AGI is difficult, so really the friendliness problem and the
question of how to make them creative (etc) already has to be confronted
and solved before we create the first one. Creating multiple AGIs would
then be an afterthought, rather than a solution to those problems.
If, on the other hand, you are talking about the R&D process that will
go on during the creation of the first AGI, then I completely agree with
you: we need to experiment with a range of mechanisms in order to find
out how they behave (and that is very much part of my own program of
research). But these will not be free-ranging AGIs that are allowed to
evolve and interact in the real world. That would be very different
from simply allowing everyone and their motheer to build a different
type of AGI, then letting them all interact and compete to see which is
the best.
Richard Loosemore
On Jan 5, 2008, at 9:52 PM, Mike Tintner wrote:
I think I've found a simple test of cog. sci.
I take the basic premise of cog. sci. to be that the human mind - and
therefore its every activity, or sequence of action - is programmed.
Eric Baum epitomises cog. sci."Baum proposes [in What Is Thought]
that underlying mind is a complex but compact program that corresponds
to the underlying structure of the world.."
As you know, I contend that that is absurd - that, yes, every human
activity - having a conversation, writing a post, making love, doing a
drawing etc - is massively "subprogrammed", containing often v. large
numbers of routines - but as a whole, each activity is a "free
composition". Those routines, along with isolated actions, are more
or less freely thrown together - "freely associated" . As a whole, our
activities are more or less "crazy" walks - I use "crazy" to mean both
structured and chaotic - and effectively self-contradictory.
(This has huge implications for AGI - you guys believe that an AGI
must be programmed for its activities, I contend that free composition
instead is essential for truly adaptive, general intelligence and is
the basis of all animal and human activities).
So how to test cog sci? I contend that the proper, *ideal* test is to
record humans' actual streams of thought about any problem - like,
say, writing an essay - and even just a minute's worth will show that,
actually, humans have major difficulties following anything like a
joined-up, rational train of thought - or any stream that looks
remotely like it could be programmed overall. (That includes more
esoteric forms of programming like random kinds). Actually, humans
follow more or less roving, crazy streams of thought - not chaotic by
any means, but not perfectly "joined up" either - more or less
free-form, a bit like free verse - somewhat structured but only loosely).
I still think that this is the proper, essential approach to studying
the connectedness, programmed or otherwise, of human thought. But it
is obviously a complicated affair - even if one could record those
streams of thought absolutely faithfully.
And science likes simple tests/ experiments - the more mathematical
and measurable the better.
So here's a simple mathematical test, which everyone can try.
"Do an abstract line drawing." (for let's say 30 secs. - on this
particular site)
Here are a few of my spontaneous masterpieces:
http://www.imagination3.com/LaunchPage?aFileType=&_nolivecache&sessionID=&message=&room_email=&[EMAIL PROTECTED]&from_name=mike
tintner&[EMAIL PROTECTED]&to_name=&aDrawingID=20080105_194101926_970043768_gbr&transcript=&_lscid=
.
http://www.imagination3.com/LaunchPage?aFileType=&_nolivecache&sessionID=&message=&room_email=&[EMAIL PROTECTED]&from_name=mike
tintner&[EMAIL PROTECTED]&to_name=&aDrawingID=20080105_194033348_926554557_gbr&transcript=&_lscid=
.
http://www.imagination3.com/LaunchPage?aFileType=&_nolivecache&sessionID=&message=&room_email=&[EMAIL PROTECTED]&from_name=mike
tintner&[EMAIL PROTECTED]&to_name=&aDrawingID=20080105_193922629_715992016_gbr&transcript=&_lscid=
.
http://www.imagination3.com/LaunchPage?aFileType=&_nolivecache&sessionID=&message=&room_email=&[EMAIL PROTECTED]&from_name=mike
tintner&[EMAIL PROTECTED]&to_name=&aDrawingID=20080105_193734879_1708083161_gbr&transcript=&_lscid=
.
The beauty of this site is that it does indeed record the actual
stream of thought/ drawing - and not just the end result. (It would be
v. interesting to see many other people's tests).
Now you guys are mathematicians - I contend that those drawings are
indeed crazy, spontaneous, free compositions - they have themes and
patterns in parts and are by no means entirely random, but they are
certainly not patterned or programmed overall either. Can you find an
overall pattern or program to any of them - let alone a program that
underlies ALL of them? Or, if you prefer, can you find a suite of
programs?
(I guess a more formal way of expressing the test is that on any given
page, it is possible to draw an infinite number of line drawings which
are a) structured b) chaotic c) crazy (mixtures of both) - and, in
principle, programmed or non-programmed. And to assert that human
activities are programmed is, in the final analysis, to assert that
there is no such thing as a crazy set of lines. But please comment).
What this test shows, I believe, is the bleeding obvious - humans can
and do produce truly spontaneous,crazy, nonprogrammed,ad hoc,
unplanned sequences of action. Well, it should be obvious but many of
you guys will fight to the death to defy the obvious. So one needs a
simple test.
It's a considerable historical irony that "painting by numbers" was
born very roughly at the same time as AI/ cog sci , c. 1950.
Cog sci. is the view that we live - paint, eat, copulate, talk, etc.
- by numbers. That view is wrong. We live, paint etc. by free
composition. (And we find both our own and nature's created forms
"beautiful" or "ugly" precisely because only a relative few are highly
structured/ effectively programmed while the majority are much wilder
and a little too freely composed).
P.S. I welcome any proper mathematical formulations of the idea behind
the test and also any suggested variations on the test.
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?&b26593
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?&;
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=8660244&id_secret=82362334-c74973
