[JOIN] Chris Paget

From: Chris Paget (ivegotta@tombom.co.uk)
Date: Mon Aug 22 2005 - 18:34:16 MDT


Hi all,

I've been pointed here by a couple of folks over at SIAI, since I have
some ideas about AGI - somewhat controversial ones by all counts. So,
by way of a join post, I'll copy and paste the relevant bits from the SI
list...

A quick disclaimer: Today was the first time I even heard the phrase
"technological singularity", although I've been exploring the idea for
some time now. I'm definitely not "up" on the terminology, so apologies
in advance for any misnomers :)

Not-quite-first, an introduction. My name is Chris Paget, I'm a
27-year-old security consultant who lives and works in London. I'm
married (for almost a year now) to an american, Erin, and we're in the
process of applying for a spousal visa so we can move back to
Pennsylvania (where Erin is from). I've been programming since I was 3
years old, and a penetration tester for about 4 years now, the last 3 of
which have been with NGS Software; if you're into security you will have
probably heard of me from security.tombom.co.uk/shatter.html, which I
published in August 2002. I'll mention that I studied Computer Science
at Cambridge Uni (here in the UK) but if you want the rest of my
(probably rather boring) history feel free to ask me :)

Now to the interesting stuff. The idea that I have been working on
apparently fits into the Wikipedia definition of "strong" AI - I know
very little about neural nets, genetic algorithms and their ilk, and
don't really want to know - it seems like a blind alley to me. Too much
research time is, IMHO, currently spent on AI as a system of complexity;
it seems like the current thinking is that if you add sufficient
complexity to a system it can "appear" intelligent, maybe even enough to
pass a turing test. I believe this is wrong; my ideas are focussed
around intelligence as an emergent system based on some relatively
simple rules.

The stuff I've had time to read on the SIAI website appears to be fairly
high-level; they've certainly solved some interesting problems before
they've even arisen, using some clever reasoning. However, I disagree
with the statement that "...the search for a single essence of
intelligence lies at the center of AI's previous failures". I believe
the problem is quite the opposite; we have not yet looked hard enough at
the overall problem to realise that the most complex problem (as we see
it) is actually the simplest - that of how to integrate emotion into an
artificial mind.

My thinkings have largely focussed upon this problem, with some
startling conclusions. I believe that emotion is a fundamental
component of intelligence, that the two are inextricably linked, and
that it may even be the case that they cannot exist without each other.
  I won't go into too much detail (just in case I'm emailing a black
hole here), but I believe that a simple system, constructed around some
very basic principles (I refer to it as emotional mechanics) can be
emergent, and that the emerging properties and behaviours are what we
would classify as intelligence.

I'm looking at the problem from a much lower level than what I've read
on the SIAI site; they've solved a lot of the high-level problems I've
thought of (and many that I hadn't), but I'm more interested in the guts
of the system - actually translating it into code that can be written.
I'm on the verge of a simple implementation, essentially integrating
emotions into an intelligent tic-tac-toe system. I haven't gotten
further than the planning yet, but it's already proving an interesting
exercise.

Anyway, I think I've rambled enough for what was intended to be a brief
introduction, so I'll stop here. Please feel free to ask for more
detail on anything I've mentioned in this mail; I'm happy to discuss
anything. Be warned though, if you do wish to ask about my visa
application the response you get may be somewhat tedious and boring - US
visas are not fun to apply for :(

Cheers,

Chris

-- 
Chris Paget
ivegotta@tombom.co.uk


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:52 MDT