AI and projects forgotten
Monday, January 19th, 2009So I started seeing an analyst a couple weeks ago. That’s good news.
Now, in two 50 minute sessions only so much can happen.
BUT one of the solid effects of having taken definitive steps to right myself (yeah yeah I know “oh, as if you weren’t right enough already” ;) is that on the back burners of my mind, I’ve been playing the recapitulation game with myself. I’ve been looking at notebooks 1/4 full of project ideas, things I wanted to noodle around with, ideas I’ve had from brainstorms to mild humidity. In evaluating which of those I want to turn my attention (albeit fractured) on I keep coming back to the idea of making something smart.
Most people, when they hear “AI” think of two things now (fortunately.) They think of the “indistinguishable from human” goal, and they think of some nebulous promises of software from 10 to 20 years ago. Used to be it was just the terminator that came to mind. Hell, I call it progress.
But what’s important about the progress towards the first type of AI is the APPEARANCE of intelligence. There are hosts of ways, usually involving some kind of fuzzy logic (and I use that term more broadly than it’s technical definition.)
Well I’ve always wanted to poke around in that world. But I never quite knew where to start (ah the failings of having skipped out on my edumacation.)
Corewars frankly is just too hard a place to start. Those people are insane. There are a number of other ‘inspired by’ type frameworks I seem to remember. Design your own tank and let your bot loose to compete with others, etc.
There’s what I’m assuming is the high end and that’s taking video games and building add-on characters (NPCs) that would have complex behaviors like “knowing how to take cover when being shot at” and “running for help.” I’d always heard that those were (with some hackery) able to be built in to certain FPSs. Sounds juicy. But I think starting small would be a better tactic.
Then there was a different category. Sort of an evolutionary petri dish where there was food, obstacles, predators, etc. and the “guys” would have simple little behaviors describing how they’d move towards food, interact with each other, etc. It’s a bit tough to describe if you don’t know what I mean (I’m guessing.) When you run these things it looks like a multi-colored Conway Life field. In the large, you’d be able to watch as patterns and group behaviors seemed to emerge.
I’d had it in my mind to come up with one of these things from scratch, which seems like an awful lot of fun. But I’m going to look around the toobz and see if I can’t find something that’s done most of the up front work.
Tough to know how to google this stuff. I’m probably going to have to play the “Wikipedia Free Association” game to find concrete resources.
UPDATE: Happens every time I claim something’s hard to find. Zooland and ALife’s link page
Meh. oh, nevermind. There’s just too much out there. It’s positively discouraging.

