Re: Strong AI Thesis (No Chinese room, I promise)

Don Geddis <don@xxxxxxxxxx> wrote:
curt@xxxxxxxx (Curt Welch) wrote on 18 Mar 2006 10:2:

There's a difference between the definition of words, and assertions made
using those words. In order to communicate, we need to agree on what
words mean. It doesn't really matter what we decide; whatever different
words mean, we can then debate claims/statements using the words.

My objection was that you are using the word "existence" differently from
most people. This isn't a claim about the world. It can't possibly be
"right" or "wrong", like the sun-centered theory.

Well, if you think we are free to define a word like existence any way we
want and ignore the reality of what can and can not exist in this universe
when we pick the definition we want to use, then you have missed the entire
point of my huge reply to you.

It's not "what's the harm". It's that your claim that math is physical
is false.

You are 100% correct that the prevailing social consensus on the correct
use of the words "math" and "physical" are just as you say, and not as I
say.

But, this is EXACTLY the same problem as the fact that the prevailing
social consensus of the correct use of the words "earth is the center of
the universe" was wrong even though that's what everyone thought was the
right way to use those words.

Just because everyone likes to use words like "existence" and "physical"
and "math" a given way, doesn't mean that use of the words is the most
useful way to describe our reality. It is WRONG, because it's a very
confused view of reality.

If you're making the absurd claim that the statement "Pi is
transcendental" is physical, then the burden is on you to prove it.

I'm making the claim that everything that exists is physical, and
everything that isn't physical, doesn't exist in any form. I'm making
claims about the truth of existence, not what does exists.

In the above, the first thing that exists is the words you wrote. They
exist in a lot of places now.

At the next level of abstraction, what exists is your understanding of what
those words are referring to and what they mean. That too is physical in
the form of hardware in your brain that is used to process all things
related to your understanding of those words.

What also exists is millions of other brains that have hardware similar to
you, which also could understand your words, and which understand the
concepts about those words (however all this hardware is slightly different
and some people which understand the words have a very different
understanding from you - just because they either think about it
differently or they were just taught wrong etc.).

At the next level of abstraction, what exists is hardware in your brain,
that represents the idea that the culture shares a common view of these
concepts. But, the common view which is shared by the culture is like the
average citizen, it doesn't exist. Only your understanding of the concept
of the average citizen, and of the "knowledge of the culture" actually
exists.

There are continuing levels of abstraction related to this that also exist
as as more hardware in the brains of some people.

But, in the end, everything about the idea that "Pi is transcendental" that
exists, is physical.

And try to do it without resorting to mind/body discussions. That has
nothing to do with our current disagreement.

Easy enough.

I've asked you specific questions, and all I request is that you answer
the specific questions, without all your multitude of asides and red
herrings.

I thought I answered ever one of your questions and threw in a few hundred
answers you didn't ask. If I somehow didn't mange to answer a question,
just ask it again. If you didn't agree with my answer, don't waste your
time asking me to to repeat it. I won't have a different answer the second
time.

You have missed the key point.

I thought I gave you more than enough that you would no be able to answer
these questions for yourself. But I guess not, so I showed above what was
physical and what existed about your Pi statement, and I'll answer below
what exists, and what doesn't exist, about your you quicksort ideas.

An instance of the quicksort algorithm
may indeed exist inside some particular brain(s) at some particular point
in time. But when I prove a property about "the quicksort algorithm",
such as its worst-case running time, the object I'm discussing is NOT a
particular set of neurons in a particular brain.

Instead, it is all instances of the algorithm, in anyone's brain, or
computer, past, present, or future. That's the set of objects that I'm
constraining. That set is an abstract idea. Not something physical.

The "abstract idea" you are making reference to exists as physical hardware
in your brain. It exists as physical hardware in the brains of everyone
who is able to understand that abstract idea. The idea is very physical.
If I were to remove the neurons from your brain that creates that abstract
idea for you, you would no longer have, or be able to understand, that
abstract idea. The abstract idea would be gone from you completely. You
might be able to learn it again by reconfiguring other neurons to once
again create the missing hardware. But unless you could do that, the
abstract idea would physically be gone from your brain.

It's all very physical.

Pi was transcendental before anyone had proved it,

This is exactly the problem I point to when saying it's important to get
the concept of existence right. You just got it very wrong - even though
you are correctly using those words according to accepted social convention
- good for you for following your education - bad for you if what you think
you said above is some true fact about the universe. IT isn't and that's
what you just got very wrong.

The concept of Pi didn't exist until someones brain was first physically
wired in a way to represent the concept.

There's no such thing as a perfect circle in the universe, so you can't
even argue that perfect circles existed before someone figured out the idea
of Pi. Perfect circles don't exist, never have, and never will. Only the
hardware in the brain that represents the idea of perfect circle exists in
many people. And the idea of Pi is an idea created by more hardware, which
makes use of the hardware which creates the concept of perfect circles.

Your insistence that "everything" is physical is not only silly, it is
counter-productive and misleading.

Well, lets See, I know what actually exists and doesn't exist in the
universe. Where as you are making statements about existence that imply
things about what exists in the universe which are flat out wrong.

It would lead people to expect false
things, such as that Pi might cease to be transcendental if only all
people thinking about it were killed (and perhaps all circles in the
universe also eliminated).

Perfect circles don't exist. That's error two you just made. And you
think you actually understand what exists? I don't think so.

If I take the neurons that create the idea of Pi out of your brain, do you
honestly think that Pi would exist for you at that point? It wouldn't. If
I remove these neurons from the brains of all people that know what Pi was,
Pi would stop existing. It would be gone from the earth.

Likewise, if I restructured the brains of everyone on earth so they no
longer knew the concept of "transcendental" then Pi WOULD stop being
transcendental at that instant.

I know these ideas go against accept social conventions of how to talk
about these things. But when you use excepted social convention to talk
about these things, you have already made about 4 errors of existence. And
that's exactly the problem I'm running on about.

That is wrong. The property of Pi is a piece of abstract, non-physical
math, and it isn't dependent upon any specific physical instantiation in
order to remain true.

The concept of truth is yet another concept that exists only in the brains
of the people that understand it. And once again, remove the brains, and
truth itself will vanish from existence.

Ok, I'm going to diverge just a little to explain why we make all these
errors of existence in our language. It's because back when the language
was being created, people had no way to know what gave them these powers of
reasoning. The didn't even know they had a brain, let alone have any clue
about what it did for them. The "mind" was the word made up as the name of
the device that had all these powers of thought and memory and reasoning
and perception.

Because we didn't understand how the mind existed, and when it started to
exist and when it stopped existing, we also didn't know anything about when
idea of the mind started to exist, or stopped existing. This is how we
ended up playing so fast and loose with the concept of existence in the
realm of mental events. Not only that, they always has the mind of God to
fall back on a the mind to rule all minds. So if there was any question
about where an idea like Pi existed after we died, they could always say it
exists in the mind of God forever.

But those days are long since past now, and we know exactly where all this
stuff comes from, even if we still don't know the complete answer to how
the brain does it. And it's high time to fix all these loose concepts of
existence - at least fix it so people understand the true reality of what
exists in the universe even if we don't try to change language conventions
right away.


When the Behaviorists from the 50's figured out how simple the mind was
- discovering it was just a reinforcement learning machine - few people
believed them.

Whether true or not, this is at best misleading. Especially the "just"
word.

Yeah, this is an opinion of mine which is not easily supported. If you
don't believe it, there's no point in going further.

Yes, the mind is all physical. But it is far from a simple
undifferentiated mass. There is clearly enormous algorithmic complexity
inside.

Yeah, but I believe all the interesting algorithmic complexity of human
intelligent behavior can be created by a very simple reinforcement learning
machine. So, to create AI, we don't have to build all that complexity into
a machine. We just need to build a fairly simple learning algorithm that
can create all that complexity on it's own in a normal lifetime of learning
(not to be confused with 4 billion years of evolution learning). But like
I said, this is just my opinion and not one I can support until I make such
a "simple" reinforcement learning machine actually work and produce human
level behavior.

The mind is in fact, not what the brain does, but it is the brain.

The mind is "what the brain does" in the same sense that my PC computer
is "doing" Microsoft Word. They are simply different levels of
abstraction, talking about the same purely-physical thing.

But trying to describe minds using only physical neurons is as silly as
trying to debate Apple's Aqua desktop vs MS's Windows XP desktop, but
limiting your vocabulary to only transistors. It isn't exactly wrong,
but it's very misleading and so unhelpful that you'll never make progress
on the interesting issues if you restrict yourself so much.

Well, that's true if I'm wrong about what can be done with a reinforcement
learning machine. But I don't happen to believe I'm wrong, which would
mean anyone not talking about the transistors, is working at the wrong
level of abstraction. I do honestly believe that almost ALL AI research in
the last 50 years has made exactly that error - they were working at the
wrong level of abstraction.

Running with your transistor analogy, I believe the correct, and only way,
to correctly solve the AI "hard" problem, is to develop the right type of
transistor, that would allow you to throw 100 billion of them into a pot,
and then step back and let them self organize. No other solution of
hand-wiring transistors would have any hope of equaling human intelligence
because the "circuits" to create human level behavior are way beyond the
understanding of any human. We can hand-wire small parts of human behavior
(chess games, chat bots, etc), but we could never touch full human level
behavior with that approach. The small (by comparison to real humans)
hand-wired projects can be good and useful technologies on there own, but
we have no hope of reaching human levels of intelligence with that
approach. We must figure out how to build self organizing machines and
that means we are building a reinforcement learning machine.

I've been looking for those "magic transistors" for about 30 years now.
It's been slow going, and it's just a hobby, but I've made some good
progress lately and actually have magic transistors that have what I see as
very interesting properties. Other people aren't as excited about them as
I am and they seem to wonder why I gave up such a good cow for them. :)

But, as I said, this is just my (very strong) belief. There's no way to
prove it's valid other than to find those "magic" transistors (or prove
that the human brain is made up of magic transistors).

--
Curt Welch http://CurtWelch.Com/
curt@xxxxxxxx http://NewsReader.Com/
.

Relevant Pages

  • Re: GOD=G_uv FOR IQs OVER 150
    ... >> intelligence is related to brain size has been disproven. ... We came to appreciate that existence. ... >Your attempt to argue against a scientifc discovery using ... and therefore objectivity in science is a myth (because ...
    (sci.physics.relativity)
  • Re: Open letter to the relativity NG from God
    ... > the existence of God depends on our brain because ... So God has no existence outside of our brains? ... So your obligation is to review each and every ...
    (sci.physics.relativity)
  • Re: GOD=G_uv A CHRISTMAS MESSAGE
    ... conclude that the brain "produces reality". ... we know it" did NOT exist before our species (brain) came into existence.. ... and that RELIGION has been central to human survival for thousands ...
    (sci.physics.relativity)
  • Re: consciousness
    ... >> Icarus wrote: ... accesses some other parts of the brain that are needed to produce the ... Of course, this "I" whatever it may be, was brought into existence by ... Regarding randomness, there is no consensus on the philosophical ...
    (uk.philosophy.humanism)
  • Re: Machine takeover?
    ... closely comparable to transistors than neurons are. ... The retina communicates to the brain at about 10 million bits per second. ... make that "an Ethernet connection from 1990". ... you can even buy Ethernet hardware that does less than 100 Mbit today and 1 ...
    (comp.ai.philosophy)