[Avodah] Sephardi-ism: some food for thought

[Micha Berger]

On Fri, Dec 12, 2008 at 03:13:37PM -0500, Yitzhak Grossman wrote:
:> And if the process needs real AI, doesn't that mean it requires
:> subjectivity? Isn't the difference between real software and any
:> hypothetical "real AI" is that intelligence has an "I", a first person
:> viewpoint, a possibility of subjectivity? If you /need/ real intelligence,
:> doesn't that mean there is a subjective component?

: This is exactly the point I meant to reject, or at least question.  Do
: automated theorem provers do mathematics as well as actual
: mathematicians?  Is the mathematician's advantage his subjectivity, or
: his superior intellectual-cognitive sophistication?

Well, is "superior intellectual-coginitive sophistication" different
than just having a better algorithm than anything we have today?

To put it another way, I presumed a dichotomy:

Real AI (if possible) would have a first person perspective and therefore
be capable of subjectivity. That's "intelligence", artificial or not.
And the only means of being both non-random and non-deterministic.

Otherwise, it's an algorithm. And deterministic -- or include random
components (caused by "dice rolling" or timing oddities).

Yes, I am arguing that we want halakhah to be human not merely because
humans are better at determinism.

RRW said that he would prefer an objective system with more deterministic
results. I'm arguing that the ideal is a subjective system that includes
free will -- and thus intelligence, not algorithm of any complexity.

If someday we can create a much better computer and much better software
and do what I suggested without real intelligence, ie using algorithms
that give deteministic results that can be objectively detemined as
"right", I would still consider this inferior to a creative partnership
with the A-lmighty.

FWIW, the Mathemetician's greater advantage is probably his ability to go
beyond the algorithmic manipulation of syntactic symbols and understand
semantics. This lets him leap over things that would take an infinite
regress in software. The non-ending algorithm can be aborted because a
person can understand where things are headed. This in turn has to do
with being able to watch one's own thoughts in a self-referential loop.
All of which boil down to subjectivity. But now we're ready to start a
thread about the nature of AI -- if it could be made on topic for Avodah.

Tir'u baTov!
-Micha

-- 
Micha Berger             One doesn't learn mussar to be a tzaddik,
micha at aishdas.org        but to become a tzaddik.
http://www.aishdas.org                         - Rav Yisrael Salanter
Fax: (270) 514-1507