On 7/12/06, Tracy R Reed <[EMAIL PROTECTED]> wrote:

I've been reading a lot about functional programming, lisp, AI, etc in
the last few months. Learning some neat stuff.

I've been reading a book by Joseph Weizenbaum called _Computer Power
and Human Reason_, written in 1976.  Weizenbaum is the creator of
SLIP, a list-processing programming language, and ELIZA, the
psychotherapist computer program.  He was so shocked by the intimacy
people displayed toward ELIZA that he wrote a book about what
computers are not, and what we should not try to make them do.  I'm
learning some things about computers at their most fundamental level,
and he has some good arguments against the notion of computers ever
being as "intelligent" as humans.

There's a whole chapter on what models are and what theories are, how
they're different, and what role they play in the gaining of
knowledge.  This guy really has his basics down.  He has a notion of
"computer" as "information-processing machine for a certain task".
The role of the program, in this view, is to transform a universal
Turing machine (such as your PC) into the computer you want it to
imitate.  For example, there exists, conceptually, a machine that will
manipulate images in a variety of ways.  The ImageMagick group of
programmers spend their time defining the functions of this machine.
You take their specifications and give it to your universal Turing
machine.  Your machine says Okay, I can do that, and proceeds to
imitate that machine.

This alternate view is not a gimmick.  The guy was a computer
scientist at MIT writing in the early 70s.  But it's definitely not a
common view today.  He also does some mental exercises that serve to
give a better intuitive feel for what a Turing machine really is.

The book is fascinating, even if I don't agree with some of his
conclusions.  I tend to stand on the pro-AI side of the fence.

-todd


--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to