No one seems to understand the basic semiotic fact that language has no
intrinsic reference or relation to the real world WHATSOEVER.
The linguistic sign bears NO RELATION WHATSOEVER to the signified.
Big deal. We speak in labels referring to concepts. Many of us *do* fully
understand this to the extent that we think that it's not worth mentioning
again. BUT, to say that language has no relation to the real world
whatsoever is just plain ignorant. Yes, words are arbitrary labels that can
be replaced with any other label -- but the structure in which they are used
is a model of the real world.
No amount of verbal description will ever tell you what someone looks
like. But an actual picture of the person will tell you in a second.
Bull. Pictures have finite resolution (and tremendously variable quality
starting from very simple cartoons and diagrams). You are overstating your
case to the point of ridiculousness.
You all have the ILLUSION that words convey information, because your
brain automatically, normally unconsciously, makes sense of every word you
process, converting it into graphics and images to see if it does all make
sense.
Words do convey information. If they did not, then you couldn't learn
*anything* by merely listening to a professor with your eyes closed. Do you
wish to claim that this is true -- or do you wish to, at the least, agree
that your statement is a tremendous overstatement?
There is no choice about all this. You do not have an option to have a
pure language AGI - if you wish any brain to understand the world, and
draw further connections about the world, it HAS to operate with graphics
and images. Period.
Again, complete and total bull. If you wanted to argue for a
three-dimensional model of the world, I might agree with you -- but vision
is only one sense that can build and maintain/ground such a model. Do you
wish to claim that a brain that uses sonar can't be virtually as effective
as a brain with vision? And how about all those people who are blind from
birth? Do they not understand the world?
Because you can only understand what literally makes sense.
And, at last, you *finally* say something that I can agree with. But
nowhere do you support your argument that vision is the be-all and end-all
of sensory connection to the world. Sure, it's an awesome sense -- fast,
high-definition, high-bandwidth -- but it *can* be replaced by numerous
other things. A pure language AI is possible because all sensory images can
be translated to language. The downside of language is that it is
emphatically not high-definition and it slows down to an impossible degree
when it tries to be. The upside of language is that it is a tremendous form
of (admittedly lossy) compression. It is far, far easy to teach many things
with language rather than with silent pictures.
You guys are caught in a historical timewarp which is ending just about
now..
We're as aware of all of your rational arguments as you are. Visual
processing is moving ahead at a good pace and will be integrated into any
AGI *when it is computationally feasible* because it is clearly an extremely
useful sense -- but it is *NOT* a necessary precondition for intelligence
and waiting for it to become computationally feasible is *not* required.
Further, as a output of the process, language gives us numerous insights
into how we perform cognition. Vision gives us nothing since it is (almost)
strictly an input.
----- Original Message -----
From: "Mike Tintner" <[EMAIL PROTECTED]>
To: <agi@v2.listbox.com>
Sent: Monday, April 30, 2007 9:38 PM
Subject: **SPAM** Re: [agi] MONISTIC, CLOSED-ENDED AI VS PLURALISTIC,
OPEN-ENDED AGI
I should point out something amazing that has gone on here in all these
conversations re language & images.
No one seems to understand the basic semiotic fact that language has no
intrinsic reference or relation to the real world WHATSOEVER.
The linguistic sign bears NO RELATION WHATSOEVER to the signified.
There is nothing in "Tony Blair" - the letters/ word that bears any
relation to the person. You cannot extract any information about the
signified (Tony Blair) directly from language.
There is nothing in the word "cat" - the letters/ word that bears any
relation to the form of life referred to. You cannot extract any
information from that word about the thing referred to.
Symbols are ABSTRACT. Numbers included. Entirely abstract in relation to
the signified.
The only signs that bear relation to, and to some extent reflect, reality
and real things are graphics [maps/cartoons/geometry/ icons etc] and
images [photos, statues, detailed drawings, sound recordings etc.].
By extension if you wish to know what cats and dogs or opals and car
wheels have in common it is no damn use whatsoever looking at the words.
You can only draw further connections by looking at graphics and images of
the creatures or things - and it is only from those pictures that you can
observe whether they will fit into a chair, sit on a car seat, squeeze
into a door opening, fit into a pocket etc.
From graphics and images, you can extract a GREAT DEAL OF FURTHER
INFORMATION... - about the shape of the shoulders, ears, relation of ears
to mouth, expression, size of eyes in relation to face, size of animal in
relation to other animals etc. etc. and on and on.
A picture is not worth a thousand words, it is worth an INFINITY of words.
No amount of verbal description will ever tell you what someone looks
like. But an actual picture of the person will tell you in a second.
You all have the ILLUSION that words convey information, because your
brain automatically, normally unconsciously, makes sense of every word you
process, converting it into graphics and images to see if it does all make
sense. All your drawing of further connections between things - apparently
verbally - are actually dependent on your brain unconsciously literally
drawing those connections. When Mike D verbally explained that "an
elephant was a multiton quadruped that could not fit on a chair" he was
only able to do that, because his brain was unconsciously drawing the
relevant pictures - including pictures indicating what an elephant weighs
and what a normal chair will carry. The words alone tell you nothing.
There is no choice about all this. You do not have an option to have a
pure language AGI - if you wish any brain to understand the world, and
draw further connections about the world, it HAS to operate with graphics
and images. Period.
Plato's cave parable of how we are never looking at the real forms, but
only shadows on a cave wall is actually a parable of what it is like to
operate purely verbally. Plato was contemporaneous with the introduction
of alphabetic language - the first entirely abstract form of language.
Before that language was PICTOGRAPHIC - it was still connected with real
images.
You guys are caught in a historical timewarp which is ending just about
now..
We have just entered the age of multimedia - and the printed word - still
entirely abstract - has just been replaced by the screened word - the word
on the computer screen. That word can increasingly be clicked to reveal a
picture of the real thing. The connection between words and images is
being restored. And it is an extraordinary time. Because you can only
understand what literally makes sense.
And the more and more pictures we have of things, the better we will
understand them. Pictures, once expensive and laborious, have suddenly
become very cheap, and are becoming ever cheaper.
----- Original Message -----
From: "Bob Mottram" <[EMAIL PROTECTED]>
To: <agi@v2.listbox.com>
Sent: Monday, April 30, 2007 11:37 PM
Subject: Re: [agi] MONISTIC, CLOSED-ENDED AI VS PLURALISTIC, OPEN-ENDED
AGI
On 30/04/07, Mike Dougherty <[EMAIL PROTECTED]> wrote:
graphics, image, redrawn, visualizations - all indicative of a high
degree of visual-spatial thinking. I'm curious, are your own AGI
efforts are modelled on this mode of thought? I ask because I wonder
if the machine intelligence we build will "envision" concepts in an
analogous way to our own processes.
Visual imagery - the capacity to carry out visual mental
transformations - is a big aspect of what we do as humans, and there's
a lot of neural apparatus devoted to it. In its most basic form this
allows us to visually imagine objects, or ourselves or other people in
various situations. However, in a broad sense thinking is more of a
multi-modal process involving the entangling together of a rattle-bag
of information from various sensory modalities into a single conscious
percept. I think this is why philosophers are forever complaining
that computers will never appreciate "the blueness of blue", because
for a human the concept of "blueness" may include seemingly unrelated
sensory data or other miscellaneous concepts having little to do with
the specific wavelength of light hitting the retina.
It seems very likely to me that the processes involved with
interpreting visual information may also be hijacked and put to other
nefarious purposes, which have little to do with the purposes for
which they originally evolved. The processes used to transform visual
scenes could also be applied to more abstract concepts, and in my
estimation this is what artists, mathematicians, scientists, writers
and other creative thinkers do all the time.
-----
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/?&;
--
No virus found in this incoming message.
Checked by AVG Free Edition. Version: 7.5.467 / Virus Database:
269.6.2/780 - Release Date: 29/04/2007 06:30
-----
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=231415&user_secret=fabd7936
