> but if we look at it from an AI standpoint, a self-modifying code
> could be considered the root of AI programming.
Self-modifying code isn't all that special. It's a dangerous coding
practice, since it is more difficult to debug and maintain, but that
doesn't make it AI. Mostly it was devised by coders who had to cram
programs that should take 8K bytes into a 4K byte memory. It
survived those early days because it does have a certain elan, and
folks new to programming are fascinated by it.
What you want from AI is the ability to learn and change the program's
behavior based on what it learns. That can be done - and is done -
with data structures. LISP uses data structures. It isn't self-
modifying. In fact, from what LISP I remember, you can't actually
write self-modifying LISP in the usual sense. You can use LISP - and
most any language, in fact - to write a program that uses itself for
data, and so can make modifications of itself and then tell the
computer to run the modified version.
> If a program is unalterable, only going thru the motions of
> programmed options and actions, that would be a robot no matter what
> its appearance.
I think I wrote before that human brains are more hardwired than most
people think. We can learn and change our behavior within some limits,
but everyone has a point where the wiring takes over, and we can't
change. Just where we draw the line between "changed behavior based on
new information" and "programmed options" isn't terribly clear to me.
What it mostly boils down to is that people think of human intelligence
as something special. It may not turn out to be all that special, but
just more complicated than most robots with "programmed options."
mailto:[log in to unmask] Just another guy
http://hmoulding.cjb.net/ with a weird name
Send a cool gift with your E-Card