Re: determinism, freewill, chaos, and circular causality


Curt Welch wrote:
"Stephen Harris" <cyberguard-1048@xxxxxxxxx> wrote:
"Curt Welch" <curt@xxxxxxxx> wrote in message
news:20060416185725.471$D6@xxxxxxxxxxxxxxxxx
"feedbackdroid" <feedbackdroid@xxxxxxxxx> wrote:
Curt Welch wrote:

Just like noise will drive a logic gate into an unpredictable direction
if it's near the border between two attractors. The point is that
unpredictable behavior is not intelligent behavior. Intelligent
behavior happens when you eliminate unpredictable behavior and replace
it with predictable behavior.

Human behavior should not be looked at as unpredictable. It's simply
complex. There's a big difference. If intelligent behavior was in
fact unpredictable and chaotic, we would all be dead.


It is not that simple. There are designed patterns which are intelligent
and sufficiently complex that the underlying rule which generates the
pattern cannot be identified. The pattern cannot be distinguised from
some sample of truly random data.

For example, Pi is expanded to zillions of digits by a fairly simple
algorithm. Pi also passes randomness tests. The newer definition
of randomness says that randomness cannot be generated by some
rule, so Pi is not considered truly random.

Suppose a sample of a million digits of Pi's expansion is provided
(without you knowing about it) by a much older advanced and
intelligent civilization which has surpassed the limitations of human
technology for computing Pi. Could you compare that string of
a million digits to a truly random string of a million digits and then
distinguish which string originated from an intelligently designed
algorithm which generates the next digit in a completely predictable
manner?

No. There are intelligent patterns which are not distinguishable from
unpredictable patterns. There are also truly random sequences
which might intuitively appear as part of a pattern that are not.
So your observation is only true for a limited domain of all such
possible patterns. This becomes relevant when the question is asked
can a computer intelligence surpass a human intelligence and how
does a human establish the criterial to answer such a question.

Regards,
Stephen

I don't happen to agree that digits of PI are an "intelligent pattern".

Intelligent patterns are the ones that help us survive (or reach our goals)
(aka intelligent patterns of behavior). No matter how random the behavior
seems (or is), it can still be tested for it's survival advantage. And if
it helps us survive, it's intelligent behavior even if it's otherwise
random.

That makes for a simple IQ test. Drop the subject in the jungle
and if they survive they are intelligent.

If behavior that leads to survival is your criteria then the *** roach
and the scorpian will be real intelligent if the atomic bombs go off.

How about reproductive success. If someone on social services
has umpteen children and a brilliant scientist has zilch it makes
the person on social services the smart one.

Academic intelligence may not survive but it the kind that I think
AI is aiming for rather then the intuitive intelligence from which
academic intelligence evolved.

That's why I said that if human behavior was unpredictable, we would be
dead. No matter how complex our behavior is, it's happening because the
brain has selected the current class of behaviors based on past experience.
Even if the brain is currently selecting "random behavior" (which I don't
actually believe it ever does but it could have that option) that behavior
was still selected because experience indicated that it would be the "best
guess" at what to do.

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

.