Ben,
You seem to have gone off in flights of fancy.
BEN:"It's trivial to write a single, short computer program that can
generate *every possible picture* that can be displayed on a computer
screen, one after the other -- including all the curves you like to
draw.... This program would indeed use simple math equations. It
would create a digital image of every beautiful painting ever made,
and every one that ever will be made.. for example..."
No it's not "trivial" and it's never been done, and never will be done. What
on earth gives you the basis for anything you've just written? Once you
unquestioningly posit such a magical entity - an "all-shape assuming"
program - you can get totally lost in the "logical" but totally "fanciful"
consequences.
Put what I wrote below into more visual program terms -
the reality is that there are no visual programs whatsoever (autonomously
form-changing programs vs aids-to-human-artists programs) that do not have
an EXTREMELY NARROW REPERTOIRE OF VISUAL FORMS.
There are Mondrian programs that can produce endless variations on
pseudo-Mondrians - with lines and rectangles - but THAT'S ALL THEY CAN DO.
They can't suddenly mutate into producing new kinds of forms - Rothko
rectangularish forms, or Miro "blotty" forms, or Jackson Pollock "blotting
pad" forms - or any such diverse forms whether similar to an artist or not.
They can just do their lines and rectangles. They can't mutate into curves.
Whereas a human playing around with doodles can endlessly generate new
species of forms.
And if you think they can - PRODUCE ONE FUCKING EXAMPLE.
Why is this? Because there are no formulae/algorithms that can cover diverse
"species" of forms. I've often made this point before but there seems no way
it can penetrate you guys - geometry's formulae are EXTREMELY LIMITED -
they can only produce v. limited species of geometrical forms - and thus
there are and have to be thousands or millions of them - there isn't just
one geometrical formula/algorithm that can produce every geometrical form
whatsoever - triangles AND squares AND circles AND Mandelbrot curves...
No wonder you're lost if you can even entertain such a notion as you started
with here.
It's worth taking time to understand the NON-GENERATIVITY message, because
it applies to every kind of algorithmic program whatsoever - artistic,
musical, building, cooking, circuit-building....
And once you get it - and it's not hard - I will be your saviour.
************
As for the "How is creativity produced?" again you've boxed yourself into an
absurd corner.
You've started with:
"well of course creative programs are algorithmic - if he doesn't believe
that he must believe in magical creativity".
To repeat: there are no creative algorithms - that's as absurd as your quote
above. But that doesn't mean for a second that creativity is
nonmechanical/"magical"
How do you actually create your own home-made stew, or improvise your own
tune on a piano? Think visually of what you actually do, and you'll realise
those are mechanical, physically instantiable affairs.
You reach out for some foods that might be suitable, toss them into the pot,
and see what you've got. You reach out, press some keys down and see what
noises emerge. A machine can do that.
Hey that's"improvisation." Real improvisation - which you really have not
understood. Those musical programs you quoted before are merely
"permutation" programs - ditto GA's - there's no improvisation. They
permutate a given set of elements, possibly then further permutating the
resulting permutations. That's not improvisation.
With true improvisation you physically or mentally reach out and discover
"objets trouves". Found objects. Newly found objects. New elements. You
physically explore the world and bring in new elements to the mix of
whatever you're trying to produce. And there's no "prediction" involved,
just creative, adventurous trial and error - you won't know whether anything
works until you've tried it.
Your GA's are not creative because there are NO NEW ELEMENTS. They merely
play around with a GIVEN, FIXED SET OF ELEMENTS.
Life, every which way, is creative - continually incorporating new elements.
Sexual unions involve new mixtures of genes.
Everyday, Turing-test, conversations are creative - continually
incorporating new elements - which is one reason why they will always defeat
algorithmic approaches. Today you're talking about Romney-Obama, Armstrong
doping, Spain going to the ECB - and there's never been anything
formulaically like these events.
That's what it is to be a conversing human being - continually creatively
improvising and incorporating new elements into your conversation....
Algorithms are CLOSED SETS. AGI is about endlessly mixing in new elements
from the world (and your own infinite range of movement and thought) into
your courses of action.
-----Original Message-----
From: Ben Goertzel
Sent: Thursday, October 11, 2012 8:31 AM
To: AGI
Subject: Re: [agi] Behold your saviour, Ben
Mike T,
About programs to generate geometrical shapes
Let me turn your question around a bit...
It's trivial to write a single, short computer program that can
generate *every possible picture* that can be displayed on a computer
screen, one after the other -- including all the curves you like to
draw.... This program would indeed use simple math equations. It
would create a digital image of every beautiful painting ever made,
and every one that ever will be made.. for example...
The question is then how to filter down the program's output, so that
it generates only the shapes you want it to. If you have, say, 10 or
20 example shapes, then current machine learning tech can learn a
model of these 10-20 shapes, and try to create new shapes in their
same spirit...
For simple classes like circles or lines, this would work fine...
For more complex classes of shapes like, seashells or dog faces, a
simple machine learning approach won't work unless you give it
insanely many training examples. To deal with systematically
generating these more complex classes of shapes you need a more
complex and subtle AI system than anyone has created to far.
However, one could prove a theorem that: For any category of shapes
that can be shown on a computer screen, there is some computer program
that will generate all and only the shapes in that category...
The fact that we don't currently know the exact program for
generating, say, the set of all images of dog faces -- doesn't mean
that there is no such program. In fact we can prove via mathematics
that such a program exists.
Even if I knew that exact program (for generating the set of all
images of dog faces), it would be large and complex and too much to
paste into an email. And if I did so, you wouldn't know enough to
read the program anyway...
As far as creativity goes -- I think you misunderstand it. A mind is
a complex thing, with explicitly, acutely conscious aspects plus less
acutely conscious (commonly called "unconscious") aspects. Some new
creative idea may seem to the conscious mind to have popped
miraculously out of the blue. But actually it was created by the
unconscious mind via combining and abstracting from and mutating
various previously existing ideas and percepts and actions -- which
then delivered it to the conscious mind. By looking only at the
conscious image of an act of creation, you see it as more
miraculous/mysterious than it is.
-- Ben G
On Thu, Oct 11, 2012 at 1:26 PM, Mike Tintner <[email protected]>
wrote:
Mike A:
Surely you'd have to concede that there are some rules which persist
over time and are static?
Absolutely. All mathematical and logical and algorithmic systems (in
themselves) are completely, eternally non-creative, non-generative. They
are
all dead recipes with rigid rules that have never and could never produce
a
single new ingredient or element - because quite obviously they are not
designed to be creative. They are recipes with set, exclusive mixtures of
ingredients.
(This is the crux of creativity - the capacity to add new hitherto unknown
elements to a course of action or its product).
If you add new unknown elements to a recipe, the recipe collapses and
could
get v. nasty. If you allow a building algorithm that produces lego block
structures, to introduce any new building blocks - rocks, say, or chunks
of
mud, - its buildings could literally collapse. And no one tries this.
These
systems are designed to produce precisely predetermined results with
precisely predetermined mixes of known elements.
These systems are wonderful if you want to be a narrow AI cook who can
cook
one specialist dish or set of dishes. They're useless if you want to be a
creative cook, who can endlessly generate new dishes, as humans can.
Now surely you can concede that no one anywhere in the entire history of
the
world has produced a single exception to this general rule of the
non-generativity of formulaic, rulebound, set-ingredients systems? There
are
no algorithms, formulae or logics that are creative. No one has ever
produced an example here. No one ever will.... And there are zillions of
possible examples.
What we do have is the most amazing amount of logical gobbledygook that
argues how these systems might be creative - but neither a) explains how
they can introduce new elements or b) provides a single instance of a
program etc that ever has.
Nada. But an awful lot of shameful assertions that of course there are
such
systems - and of course people have produced millions of examples of them
in
the past - and how could you, Mike, be so stupid as to think there are
not -
and ROFL at you - oh absolutely ridiculous - but now, right now, the
speaker
is just too busy, you understand, to produce a single example. Oh of
course
he could produce *so many* examples, and he will, he will, but now right
now, he can't. (Basically all people who argue thus are lying gits).
If you or Ben can grasp this simple obvious truth of the non-generativity,
non-new-element-ality of formulaic, rulebound systems with set mixtures of
ingredients, I will indeed be your saviour.
What you et al are trying to maintain is a scientific, material
absurdity -
and something of which you will come to be v. v. ashamed. Produce ONE
FUCKING EXAMPLE. Or admit you can't.
P.S. And I've heard all the shit about sophisticated, evolving systems and
GA's etc - they cannot and never have introduced a single new hitherto
unknown element They have no novelty. Demonstrably. They are mindblowingly
narrow in their products except to AGI suckers who actually half believe
their own hype - and AGI is nothing but failed hype.
-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/212726-11ac2389
Modify Your Subscription:
https://www.listbox.com/member/?&;
Powered by Listbox: http://www.listbox.com
--
Ben Goertzel, PhD
http://goertzel.org
"My humanity is a constant self-overcoming" -- Friedrich Nietzsche
-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/6952829-59a2eca5
Modify Your Subscription:
https://www.listbox.com/member/?&;
Powered by Listbox: http://www.listbox.com
-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393
Modify Your Subscription:
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968
Powered by Listbox: http://www.listbox.com
