On 3 xxx -1, Marchal wrote:
> Scott D. Yelich wrote:
> >Knowledge is when predicted.
> Interesting (?). Could you elaborate a little bit?
> Bruno

Thank you for being interested.  Please understand that I do not have a
background in philosophy, logic or computer science, per se.  I am only
interested in what can work in a way that is easy to understand.

I joined the everything list because I am interested in AI. Actually,
I'm interested in what I was calling "Real AI" ... and now I find some
people are referring to it as "New AI," and this is somewhat good as
I no longer think I am alone in the direction I chose and I feel
safe unencrypting my efforts.

I have been thinking about AI as a past-time since about 1988 and I can
remembering tinkering a little long before that. My current goal has
been to implement a system that is capable of unrestrained evolution,
but evolution based on Intelligence -- not wet-ware.  This system
doesn't have to be perfect, mathematically sound or proveably complete.
It just needs to work.

To do this, I was trying to come up with my own theory of everything
that I could use to code.  If I can understand the scope of something --
I can code it.

What I see on this list is a lot of smart people who argue about the one
true way.  I'm not really interested in the one true way -- just any one
way that (is sufficient and) works.  I have been trying to gather any
little bit of insight from this list during the time that I have been
watching it.

I have come up with:
Difference is not the same, even when it's the same difference.
Difference is not the same, even if it's the same difference.

I can't decide over the difference of those two, that is -- which I
prefer.  I doesn't seem to matter too much as it all comes back to
language being meaningless at the lowest level anyway.  I have posted
this difference thing to the list and, of course, it was ripped apart. 
I have yet to get a clear response from anyone, especially people  who
think that time doesn't exist.  The reason why I have been pursuing that
so strongly is I am basing everything on the fact that "everything" 
doesn't (or can't) happen at once.

I have been able to explain everything using this quirky little ToE and
since I can code this ToE, I am quite satisfied.

To me, if there isn't difference, then everything is the same.  To go to
the opposite, right when there is difference, how can this be
determined?  It just seemed to me that when there is difference, it is
then possible to start to define things -- that is, things are no longer
all the same (ie: things are defined by difference, not by sameness --
we define things through differentiation).  The issue is how to tell
what is different?  I can't quite come up with an intelligent enough way
to explain this, but it's basically that if there isn't time, everything
is the same.  There are no two different points to use as what this list
might call observer moments, there would be only 1 observer moment --
and I'd go so far to claim that then that observer moment was really
impossible (at least from within the system).   All of this crazy and
seemingly circular musing just lead me to base my ToE on the
fundamental building block of "difference" ...

In order to code a system that has real AI, I came up with the follwing
ways to define things so I could code them:

Knowledge is when predicted.
Concepts are reduction or expansion of knowledge.
Meaning is a concept that is relative and arbitrary.
Reasoning is the ability to create meaning through explanation.
Intelligence is the ability to judge better or worse via reasoning.
Communication is to convey meaning over language, which is arbitary.
Life desires sameness, but exists only in a world of difference.
Life desires sameness, but existence is only difference.

The last two lines show the circle and how the system becomes a
perpetual-motion machine.  The idea behind those lines is that the term
on the left is what we humans tend to reference when talking about life,
learning and intelligence.  The stuff after the word "is" is how I plan
to achieve the words at the left  (the last line could be: Life
is difference, but desires sameness).

So:

differentiate - discern - store   - recall  = biological learning
approximate   - choose  - predict - compare = computational learning

I recently switched jobs, but I should be able to get back to some 
coding RSN (I hope), but this was mostly finalized by Dec 25, 2000.
Coding was started then.  To me, this is extremely exciting.

I have been coding the "biological" side of this system.  Basically,
it's an IO system based around UNIX sockets where the system can
create/destroy new "connections" itself.  This is just the way I chose
to model my ideas -- it's not the only way possible.  Anyway, I am using
these IO paths as senses.  The system then automatically "nodifies" the
inputs/outputs of these senses so that it can learn, biologically, from
its environment. This is done so that some "learning" is inherent to the
system, at least at the "biological" level -- but it is all based on 
the Difference ToE.

So, the premise is that if the system knows certain things biologically,
then perhaps that could be the foundation for a higher level.  What 
does it mean to "know" something?  Well, at the biological level,  it
means that what the system is currently trying to figure out if it
"knows" is either already stored or not.  So, this stuff is "nodified"
(ie: discerned).  However, part of nodifying/storing is that the same
sense doesn't create two nodes for the same thing, so the system will
know, at least biologically, if it has experienced this sense before.

The next level is computational.  By comparing some possibilities and if
one matches, this means that the system "knows" on a computational
level.  Example:  physics .... somehow we can predict the path of
projectile with great accuracy.  To me, this means we "know" something
(and more than likely actually "know" many things in order to do this,
but that's not the argument).  That's an advanced example, but I think
it can be reduced down to things such as syntax, grammar, etc.

The bottom line here is that, so far, I haven't found a single person
who can break my ToE or the terms as how I have them defined above.
Since I can code what I have defined above -- I am satisfied. I also use
that list to try to reduce what people are saying on this list to see if
anything new is happening, instead of the seemingly constant vicious
circle of bickering over nits.  I'd rather simply evolve bits.

Scott

ps:

So, to code an AI --

Knowledge is when predicted.
 If something can be predicted, then that "something" is "knowledge"

Concepts are reduction or expansion of knowledge.
 The system will perform certain tasks in order to predict.
 When this same "certain tasks" can be used to predict more than
  one single instance, then this is a "concept" ... that is, 
  this procedure is taken itself to be something.

Meaning is a concept that is relative and arbitrary.
 Meaning is simply a concept.  Meaning, like language, is simply
  brought about by consensus.  Meaning doesn't have to be right/correct,
  it only has to be a concept that is accepted (to the entity
  itself, or shared with other entities).

Reasoning is the ability to create meaning through explanation.
 This is also similar to predicting.  However, this uses the
  rules (logic?) of reasoning, vs the end point.  Again, reasoning
  doesn't have to be correct, it only has to be a concept.  This
  is similar to concepts, except that this is almost like the
  "act" of creating a concept.

Intelligence is the ability to judge better or worse via reasoning.
 This is the crux.  Intelligence is simply a (I forget what the CS
 people call this... I'm sure the minute I send this damn post
 I'll remember) ... way to determine when something is better or
 worse.  This is completely arbitrary because an entity can
 choose to believe that something is better or worse regardless 
 of any relation to "reality" ... but this is still intelligence,
 as the "fitness" of the entity is compared to others.

Communication is to convey meaning over language, which is arbitary.
 See Meaning.  To me, communication is simply a symbolizing of
 concepts -- that is, communication is meaning.  It's relative
 and arbitrary.  It's thhat the purpose of communication is different
 than that of meaning and concepts.

Life desires sameness, but exists only in a world of difference.
Life desires sameness, but existence is only difference.
 If there only sameness, there is nothing.  When there is
  difference, life seems to want to figure out why.  Figuring out
  why means seeking the sameness (which doesn't exist).  This
  is what people might call "will" or "purpose (of life)" ... etc.


There are, of course, other little things that need to be in place in
this system, but they are minor as compared to the overall philosophy.

The other simple things is that this system can't evolve in a vacuum.
That is, a system can not go outside its bounds, but through interacting
with its environment, the system expands its universe beyond what it
could ever hope to understand fully -- and this is enough to drive it
perpetually.

Scott


Reply via email to