Re: Kress's Probability Trilogy Q's

In Message-ID:<9t2h6399m5g5av7ce6nqia5g9vit4usp9r@xxxxxxx>,
James A. Donald <jamesd@xxxxxxxxxxx> wrote:

--
James A. Donald:
Anyone who claims to have a definition of thinking
or understanding is full of crap, and a fair bit of
Penrose's book was intended to demonstrate that.

Arthur T.
It's useless to make such an argument if you can't or
won't define "think" and "understand".

Fallacy.

Definitions can only be given in terms of something
else, so to avoid endless circles, definition must
finally rest on example. Penrose gave examples.

A definition in your sense is only useful if we can
define something poorly understood in terms of something
better understood - but all "definitions" of
consciousness work the other way around.

If we cannot figure out how the neurones in the retina
do data compression, we are a long, long, long, long way
from defining consciousness.

Absent definitions, you need at least a decision procedure.
It's useless to try to determine whether something does or can
exist if you can't recognize it when it jumps up in front of you.

Turing suggested that if you can't tell whether responses to
your questions are coming from a human or not, whatever is giving
those responses must be considered as being humanly intelligent.
This is a very high bar, but it's hard to argue with.

The Chinese Room meets Turing's test. Why does Penrose feel
it's not intelligent? Basically, because when you analyze it, the
idea is "obviously" preposterous.

Note that that is the same "obviously" as in:
James A. Donald
A human can see that the rules of simple arithmetic
must be consistent. An algorithm cannot.

Just because people have a feel for what is true, doesn't
make them correct. For hundreds of years, people thought that
Euclid's fifth postulate was "obviously" correct and discounted
every foray into non-Euclidean geometry. They tried and tried to
find the fallacies in them because they "knew" they couldn't be
consistent.

Basically, what we "know" non-algorithmically is merely known
empirically. From that, it is easily possible to draw wrong
conclusions. Even if we draw the right conclusions, we have no
way of knowing that we're right.



--
Arthur T. - ar23hur "at" intergate "dot" com
Looking for a z/OS (IBM mainframe) systems programmer position
.

Relevant Pages

  • Re: Kresss Probability Trilogy Qs
    ... An algorithm cannot. ... first-order arithmetic is consistent, ... http://www.jim.com/ James A. Donald ...
    (rec.arts.sf.science)
  • Re: The downsides of pseudo-historical fiction
    ... James A. Donald: ... conspicuous counter stereotypical virtue, and no ... thug who smacked her around. ...
    (rec.arts.sf.composition)
  • Re: WorldCon 2006
    ... James A. Donald: ... "Keith F. Lynch" ...
    (rec.arts.sf.fandom)
  • Re: Bush Strums as New Orleans drowns
    ... >James A. Donald: ... I have essentially zero interest in finding out. ... is not a fascist dictator, I don't see any useful purpose being served ...
    (rec.arts.sf.fandom)
  • Re: Need help with expanding energy dome
    ... James A. Donald wrote: ... Devices like remote cameras can tell us what is or is not happening. ... And if you still wish to argue that there are areas in which incidents ...
    (rec.arts.sf.composition)
Loading