Re: A request for information please.

N wrote:
On 17 May, 22:16, Stephen Harris <cyberguard-1...@xxxxxxxxx> wrote:

Sergio Navega used to be a regular contributor to c.a.p. In this
paper he makes the point that the person who combines education
or reading along with practical experience is going to produce a
foundation for a deeper philosophical perspective than a person
who lacks such experience. He relates this to how an AI "ought"
to function.

http://www.intelliwise.com/reports/paper1.htm

"Later in this paper I will present some speculations on a
different point of view: that knowledge representation and
inference should be almost indistinguishable from one another
and the intelligence that a system presents is the result of
a _simultaneous_ _growth_ in its capacity to _represent_ and
_reason_. This approach seems to indicate the direction of
connectionist systems, in which this integration is achieved
naturally. In fact, several connectionists see no other way
of obtaining intelligence by other mechanisms. However, I
will suggest that it is possible to find non-connectionist
mechanisms that exhibit similar behavior, with the added
benefit of being more "economic" in terms of hardware
requirements(1). ... Thus, it seems unreasonable to ascribe
our distinguished intelligence just to better perceptual
capacity. But the fact is that we are more intelligent than
any other animal. There is, certainly, something more. What is
it? Is it language?
Could language explain our intelligence?

I wish I'd take my camera this morning, cos theres a
storm brewing, coming in over the hills. That kind of
light you get where the green trees kinda vibrate and
twinkle, the yellow in them contrast against the deep
blue violet of the cloud. It'll always be windy at certain
times in the moons phases, and the sea birds will come
in from the coast, etc.

Now if I were to talk of atmospheric pressure and took
barometer readings, did a stats chart instead, I may well
conclude, that for some unknown yet-to-be established
purpose or reason, theres always going to be more gulls
inland on a certain day. What would be wrong would be my
saying that when I draw up a chart, my very act of
performing the calculations and drawing the chart DOES
bring the gulls inland. In much the same way, how could it
be possible to invoke some kinda consciousness into any
artificial construction? and for what purposes?

A chart and plan or design can't do other things,
it can't frinstance tell you that air pressure alone unequivocally
caused a headache, it can't tell you it has a headache, as
you would be blameless of causing a headache in the cloud
because you drew a graph.
You can go through all the senses, and conclude the same.

what you can do is make an inference, jog somebody's
memory in one area of your sensory experience to jog another
to conclude, in some small error, that a particular action,scene,
value etc has been causal, but you have to judge if the 'error'
margin can be taken as factual, as realism?


There is one theory (Jeff Hawkins) that memory stores the core
elements of some experience, but not the exact experience, so
that each time one recalls a memory it is newly reconstructed
slightly differently than the original. Also it is likely that
memory has associations. For instance, I saw a trailer for a
movie that had a famous actor in it, but I couldn't remember
his name. But I could remember his movies, Godfather, and that
triggered Last Tango in Paris. Then came his name, Marlon Brando.

So I take my photograph, I tweek the composition and put it
online, then someone tells me they like it but they can't
live with it! something to do with "making my room feel and
look like a dump'n'tip and gives me a strange desire to take
an asprin, and put up an umbrella after locking all the windows!!" :)
more usually it'd be "I don't know why, but it makes me feel
like....its weird"


I am interested in logical fallacies. One of them is called
False Cause. Suppose a person observes two events, A and B.
The person might think A causes B when actually the events are correlated, A and B are both effects of some common cause C,
which causes both events A and B to occur, but it is possible
for the observer to make a mistake because A appears to cause B.

Correct me if I am wrong, I think most of what you are talking
about in the above paragraphs falls under the category called
"pattern recognition" (PR) which covers a vast range of human
intelligence and is a (the) major goal of AI. PR starts when a
baby learns how to make its body interact with the environment.
Or to borrow a famous example, 'crying will cause mother to come
to the baby, but crying will not cause a butterfly glimpsed on
the other side of a window, return and fly by the window again
so that the baby can see it again.

If language were on the center of our thinking, it obviously
would be all we need to transfer any kind of knowledge from
one person to another. This, indeed, seems obvious for a lot
of people. But this is not what happens in practice. Language's
most problematic characteristic is its difficulty in conveying
sensory experiences.

same difference, you have 40,000 different languages
worldwide, to say 'doggie' and there are howmany-billions
of people who'll individually have a unique and personal
understanding? What's remarkable is we all share the same
turf, Or maybe, what is remarkable to me is we share common
bonds together with a diversity and capabilities to adapt.
Is it the adaptability or flexibility that makes evolutionary
intelligences (if you like) so precious?


I imagine so. One definition of intelligence is the capability
of solving novel problems. New problems show up in adapting to
the environment even today. As a species we can breed with each
other. Evolution has shaped our brains so that all healthy humans
have brains structures which perceive reality in the same way,
but not identically. Solving survival problems gets the offspring
to reproductive age so that their genes get fixed in the gene pool.
So having the amount of intelligence to survive longer, especially
in the prehistoric days, was saved into the gene pool by a selective
process which made more and more future offspring, more intelligent.

Have you tried to teach somebody to ride a bicycle using just
words? No, it's not possible. Without experimentation, without
feeling the difficulty to balance, the problem of coordinating
hands, feet, etc, one would not learn how to ride a bicycle.
The same happens when one is learning how to drive a car: you
can't do it "by the book", you have to get inside one and
exercise the controls. Make errors and learn to correct them.
Ditto for flying a plane. This knowledge is not acquirable just by
reading. This knowledge is not transferable through language alone.

My cat just gave birth moments before/around the time a
visitor called. She found the safest place (my lap) and
purred like crazy just to be terrified the moment at giving
birth. There you are riding your bike going your way and
someone jumps out in front of you risking life or throwing
you into complete confusion. Now just ride the same bike
twice.

So I'd say we'd have to separate language with knowledge,
and I think the 'language' meant here is 'text' or symbols
that are abstracted sub units. Alone they mean nothing,
but as part of a chain in some afferent communication to
be recognised within the system and allotted to different
areas for 'processing' in light of previous experience and
knowledge. If some foundational knowledge is a priori,
what's a feeling? is it an instict? can you design a machine
with instincts!


It is pretty likely that spoken language existed before there
was written language, so there was a time when there were no
external text or symbols, until written language was invented.

There was probably a time when human spoken language wasn't
much advanced over the 100 or so sounds and gestures that
chimps have. Internally, one could probably call the physical
patterns, the neural synapses, which associate some mutually
agreed upon sounds with physical events (danger=snake) symbolic.

Programming code, letters or images, are symbolic. That code can
have a higher or more human comprehensible expression. Or some
nerds can actually write in machine code, 0s and 1s, which are
like electrical on/off switches. When the computer operates, the
switches produce a series of electro-magnetic states (em) and it
is these em states which I don't consider symbolic, not in the
same sense one thinks of a Greek lambda symbol or oo (infinity).

Likewise when I hear words, I don't jumble the letters around in
my mind to produce some understanding and then reply. What I
think happens is that the words are converted to meanings which
are stored in synaptic patterns. The symbolic words get converted
to meanings, thought occurs, and then some _meaning_ gets
translated back into words as symbols. Well, they do call this
symbol manipulation. Perhaps it is my misinterpretation, but you
wrote something that made me think you thought of the symbols
as more concrete rather than at this level of abstraction.

Instincts! The Computationalists conjecture that instincts are
not necessary to accomplish thinking, as a computable process
which uses effective procedures to manipulate symbols.

Humans have a sense of "Iness" or ego. Instincts operate on the
ego and they are also genetically acquired so quite similar.
Instincts operating on the ego has a lot to do with organizing
what the human does and what choices she makes. But thinking in a
logical manner can probably be distinguished from intentionality,
(beliefs and desires) and the goals humans set for themselves.

It seems that a lot of intelligent capabilities can be simulated
without instincts, like the Grand Master chess playing program.
How far can these capabilities be extended? Do they include a
program that can translate poetry from one natural language into
another (languages don't always have exactly identical expressions
available going from the original to the translated language) and
then back again without garbling the poetry?

Perhaps there is a level of complexity in which simulations of
human-level intelligent behavior by purely mechanical means is
not possible without possessing consciousness or a mind. In the
book "I Am a Strange Loop" by Douglas Hofstadter, he says that
he thinks a human level AI will need a "self symbol" ... then
does this AI ego require instinct-like motivations in order to
exist and operate?

I store my experiences as 'feelings' hypothetically like knowledge
was compressed data. The data file 'module' has its own class,
colour,shape which are far easier to throw about and link together
without decompressing the entire contents of the folder. I don't
actually believe we think this way, but its sort of easy to imagine
how an artificial mind could work with higher reasoning problems
in that way.

As we suggested in the previous section, human level intelligence
demands human level sensory perception. But artificial intelligence
(in the same way that an airplane is an artificial bird) may
eventually be constructed even without all those sensory equipments.
It will not be able to perform like a human being, it will never
understand some of our more "sensory" analogies, it will never
understand our world the way we do. But my bet is that it will be
able to do a lot of useful things, like a reasonable level of natural
language understanding, an intelligent (and comprehensible) teaching
tutor, a creative (and amazing) assistant to the scientist, a (serious)
centralizer of corporate knowledge, a worthwhile technical help desk
(with infinite patience), etc. These are subsets of the old dreams of
AI. And all of these applications seem to fundamentally require the
basic principles that stand behind intelligence."

SH: Philosophy still seems worthwhile doing :-) This last paragraph
runs against the current AI tide of claiming that for an AI to work,
it must have input from external sensory data. Sergio says not so,
although less than fully featured, still quite useful, and perhaps
not requiring that a mind be instantiated to obtain that usefulness.

I looked up Wolframs competition and the Turing machine,
and that great music link.
I mean you can't really get any shorter than 'Yes'n'No!'
I don't know what it was, maybe it was the idea of the infinite tape,
and connections and stuff, a thought occurred to me about human
joints,
you know like, hinge, ball and socket, semi-moveable and all the
different ways of making connections in 3D.

How does the puppet master control the strings at the right time&place?



There is another debate about whether the brain produces the mind, or
if the mind has another source of origin (called Dualism). Another
belief is call Panpsychism, which thinks everything in the universe
might have some degree of consciousness, animate or inanimate.

How do you tell when your super-duper AI program actually rises to
the occasion and instantiates a mind? People haven't been able to
come up with any better test than the Turing Test which I mentioned
before. One or a panel of judges, evaluate its output, its observed
behavior, and see if it is human-level conversationally convincing.

This quandary is sometimes known as the Measurability Problem. How
do you know that Panpsychism is false based on some other test than
whether it moves or not, and -> your computer AI program doesn't move.

If you build your super-duper AI program how do you know it has
consciousness and a rock doesn't have any. Your AI program could
just be simulating consciousness or mind by making clever responses
like the GM chess playing program which can't explain why it made
its move, which humans can. Does the electricity tell the story.

"Measurability: With a purely contemplative position, the agent may grow intelligence but nobody external to it will be able to evaluate it. Evaluation is done by observing the agent interacting with the rest of the world. Let's take a look at one extreme example: say a rock is able to alter it's internal quantum properties based on the "information" it captures from the environment (light, temperature, humidity, vibrations, etc). These internal quantum properties have a special "mechanics" to detect and organize the received patterns into rules and "theories" that can be used to give the rock predictive abilities. However, the rock does not tell us anything about this. So, for us, it keeps being a dumb ignorant rock."

The Philosophy of Mind/AI has quite a few unanswered questions including
whether the unanswered questions have answers which can eventually be
found, or if there are unanswered questions which will never have an
answer, and how do you tell the difference between them?

My favorite peculiarity of philosophy is Pi. Pi is a computable infinite
number which means there is a shorter way or input, to cause an unending
output if you have a Turing Machine. Pi passes the tests for randomness.
But Pi is not considered random even though it has no pattern because
the (AIT) definition of randomness specifies specifies the input has to
be the same length as the output (no shorter generating algorithm like
Pi has). This makes me think that there might be an algorithm which
simulates the human mind, that there could be a program which produces
the pattern of consciousness without actually being conscious, in the
same way the program for Pi can produce the pattern for randomness,
which in this case is the absence of pattern, without actually being
truly random itself (AI can act like Pi does to pattern recognition,
simulating the patterns of a mind but not truly being a Mind).

Is there some situation, possibly encryption, in which simulating
the random behavior (infinite non-pattern of digits) which Pi does,
fall short of a potential truly random number provides? Yes, for
finite random pretenders which are called pseudo-random and are
eventually cyclic. But Pi is infinite and supposed random so not
cyclic. This introduces assumptions about discreteness and states.

(i) Man is capable of only a finite number of internal (mental or physical) states

"As to (i), Webb appeals to Turing’s argument that a human who is performing calculations is capable of only finitely many internal states, otherwise some of them would be arbitrarily close and would
be confused (Webb 1980, p. 221).
This is justified in an analysis of human calculation, where the internal states must in principle be unambiguously identified by the effective procedure followed by computing humans. In his argument, Turing makes it clear that his “internal states” should be replaceable by explicit instructions.39 Since the instructions have finite length, they can only distinguish between finitely many states.

But this says nothing about the number of internal states humans are capable of outside the context of calculation. Ordinary mathematical descriptions of physical systems ascribe to them uncountably many states. There is no a priori reason to suppose that humans are
different from other physical systems in this respect. In fact, theoretical neuroscientists make extensive use of ordinary
mathematical descriptions, which ascribe to neural mechanisms uncountably many states (Dayan and Abbott 2001)."

I think self-reference ("self symbol") plays a major part in seeing
how this works out. How cells disambiguate their boundaries meaning
they do distinguish the cells of their body from aliens which is how
the immune system works, but apparently not infallibly. The reference
to "uncountably many states" means some states would have to be random
or noise, the filtering of which is another marvel of coherent thought.

This post contains some speculative content so feel free to disagree.
For "calculation" above I used 'thinking in a logical manner' earlier.

Regards,
Stephen
.