[Perl-AI] Fire Extinguisher

The group, as usual in AI, is following silly dead ends that come out of
CS, which has it's base in math. Even when doing things with the word
"genetic" in the name, they still are using math theory instead of turning
to biology. So, not having any idea of what is being done in AI makes you
fully qualified to do good work in the field. :)

--Paris

Don't assume that it has anything to do with my job. =) In fact, for
the past couple of years even Perl has had very little to do with my
job. Perl is mostly a recreation thing for me right now.
In the spirit of Socrates (and to qualify myself according to Paris
Sinclair's criteria :-), I'll also claim that I really don't know
anything about AI either. My academic background is in other stuff
(math & music), and I don't do AI professionally. Furthermore, IMO the
subject "AI" isn't really a subject which can be defined very succinctly
anyway. Several times I've heard a definition of AI that I like: "AI is
the study of any process we don't understand. Once we understand it,
it's no longer an AI topic."

Many people in "the public" think of AI as a bunch of people trying to
make a robot that can think like a human, but there's no way people can
do that until they can solve other, much smaller problems like "whether
that thing in front of the camera is an egg or a baseball". The most
famous criterion for "success" in AI is the Turing Test, but people
don't seem to be thinking about that much anymore, because it's not
clear how many people you'd have to fool in order to pass the test.
Also, the most successful Turing Test programs to date have been
decidedly un-interesting, consisting of just a bunch of set patterns of
responses to set patterns of input.

Many researchers seem to eschew the term "AI", and describe their work
more specifically by application (Machine Learning, Natural Language
Processing, Scheduling, etc.) or mechanism (Neural Networks, Bayesian
Inference, Decision Trees, etc.). Of course, these areas are neither
disjoint nor orthogonal.

Personally, I've been enjoying learning categorization methods just
because I think they're neat, and because they have broad applicability
to lots of problems. I don't mention it in the AI::Categorize docs yet,
but categorization methods can certainly be applied to problems other
than text documents - one current application is in genomics, where
people need some heavy methods to deal with massive amounts of unseen
data that they can't afford to process manually.

In my experience, many of the ideas people work with are quite simple at
their core, but can get very complicated when working out the details.
For example, a neural network is a simple concept - just connect a bunch
of nodes to each other and aassign weights to the connections, then
learn the best weights to produce a certain output - but when you try to
nail down the fuzzy stuff in that concept and actually implement things
that do their job effectively, it can take a lot of paper-reading and a
lot of trial-and-error.

I didn't mean to write such a tome, but there you have it.


------------------- -------------------
Ken Williams Last Bastion of Euclidity
ken@forum.swarthmore.edu The Math Forum

Your mother wears army boots. (Actually I used to go out with a girl whose
mother did wear army boots).

Yes, in my eternal lexicographic quest for a good definition of "AI",
/that/ is the only one I've stumbled on that strikes me as insightful.

Daniel Hillis uses roughtly that definition in an interview with him, at:
http://freshair.npr.org/dayFA.cfm?display=day&todayDate=10%2F26%2F1998
Select the Real Audio ("Listen...") link, and then fast forward to about
11:30.
Among the examples of things people called AI, and are ceasing to call AI
(because we understand it), are: voice recognition, spelling correction,
and computer vision.

(The one about AI being about "getting to do computers to do things that in
a human would require intelligence" is not quite as useful, I think.)


There's also a new(ish) sense of "AI" (which I thought /very/ strange when
I first ran into it), to mean what I remember previously being called
"strategy" -- basically the code that has things in a game acting
intelligently. I.e., you shoot at the monster, and it's supposed duck to
avoid getting hit. I'd consider "AI", in the old sense, applicable to
situations where you want the computer to be a clever opponent in a game of
strategy (chess, go, things that seem to require game-tree-like analysis
but that then need heuristics to prune the game tree, etc). But it seems a
bit of a stretch to me to apply "AI" (in the old sense) to the code that
says little more than "monsters should duck when someone in their field of
vision starts to point a gun at them; oh, and keep moving, but don't try to
walk thru solid objects".

But it's understandable how that usage (presumably) got started: graphics
and sound programmers working on commercial games needed someone to make
the monsters move and duck, and someone told them that that's what you use
AI for!
(cf http://www.gamasutra.com/features/19981120/gameai_01.htm etc)
This probably sounded plausable to them (in my experience, commercial game
programmers are happy to think that anything not centered on bitwise-AND
operations is probably AI, and therefore Not Their Problem), and they
applied the term to whatever made the monsters move, regardless of whether
it was ten dumb lines of C, or a whole complicated neural net system
written in embedded Icon that eats 30% of processor resources.

Ahwell, it's in established jargon now, thus enabling whole generations of
people who play inane games involving a first person shooter to bitch about
"this crappy AI! The monsters just keep trying to walk thru walls! AI SUCKS!".

Joe,

Most of the time when you mention AI, people roll their eyes and think of
sci fi. To a large extent it is still a somewhat discredited field in the
computer industry.

Personally, I think that it's time has come or will be coming shortly.
Computers are becoming powerful enough to begin to handle the complexity
that "intelligent design" requires. Expert systems have been around for a
long time and databases of all sorts of information are well established.
The next "breakthrough" in my humble opinion will be interactive and
learning algorithms which will reduce the high maintenance that all current
AI systems require. Perl's almost limitless flexibility with text
processing and pattern recogition make it my choice as a language to
achieve the goal of an interactive self programing algorithm.

Real world applications are search engines, web site navigational
assistants and program tutors. "Ask Jeves" is similar to what I envision,
but almost all the work of that application is done by a team of persons.
Ah, I am dreaming but I continue to believe it can be done. Time will
tell.

- Peter

Dreams are the needed foundation for reality. I too can't wait until a
search engine finally finds exactly what I want.

Joe

P.S. Norton Antivirus is the best =)

The main problem most users seem to have (and I bet you $100 this
is what any study will show) is describing what they want. When
I manage that, Google works just fine.
Didn't catch Melissa as quickly as the Russian programs (AVX)!!!

lee goddard