quote: Original post by ahw

Actually, it''s funny because for my Masters I am gonna have to prove the lecturers that Game AI is as serious as Academia AI ...

youpla :-P

Good for you! On this we agree. They may be different levels
of the same discipline (artificial intelligence) but as a guy
who makes his living doing "serious game AI" I think its just as tough and intensive as academia AI.

Eric

I read somewhere that more AI theory is implemented in the game market than in any other arena including academia. They talk and theorise a lot (and do cool stuff with robots) but games has bought more AI testbeds into reality than any other group.

Ahw, as a person about to be in the same boat as you (convincing AI lecturers that game AI is good) tell me how you do it :-).

Mike

VERY simple
You just take their bias when presenting your stuff.
A 3D action game is not a game, it''s a software, using 3D vectorial maths, matrices, it needs octree structure fr storing topologic data, it needs a good understanding of the APi you are working on, the AI needs to work a finite machine state, etc.
The idea is to use the buzzwords to designate all the stuff you could call otherwise, but that would sound like a game.
I have been doing this kind of buillshit since I started computing studies, and everytime I talked my way through it, especially those last two years, where I didn''t have any wreport to show (you know a lot of guys who get a Degree without any sort of report ? I did )

quote: So how do you distinguish between (for example) Democrats and Republicans? Being a democrat doesn''t make you more evil and a republican (it just seems that way ). Using one variable for reputation abstracts out a lot of information that can be used here. I suggest using one variable for each moral set, meaning one variable for Charity/Greed, one for Passive/Agressive, etc. and build a general reputation value from that.

Hmm. I''ve developed an AI (but the project I was coding it for fell apart) which could distinguish that sort of thing on a rudimentary level, which made for more general rules that are easier to code and maintain.

The character would have, in addition to strength and dexterity and so forth, two personality ratings; the first is "Alignment", which is a similar concept; 0 is completely neutral, though most characters would fall to one side at least a little bit. Then negative numbers would represent "evil", up to -127 (most "evil") and positive numbers up to 127 for the most "good".

As you say, this doesn''t distinguish between two "good" people who may be Democrats or Republicans and not get along. Thus, I include a second personality rating, "Karma". Karma is an unsigned, unitless number, which simply serves as a general measurement of that character''s personality; when two NPCs/etc meet, the game compares their alignments first, and if they roughly agree on alignment, then it proceeds to check Karma. The closer your Karma rating is to that of another character, the more likely you are to "get along" with that person. You could think of it as a political scale, if you like; Democrats and republicans would be somewhere around 75 and 175, and totalitarian dictators and anarchists would be at 255 or 0. The benefit of the system is that it doesn''t have to simply be political systems, it can be used as a full personality catchall and hold a combined record of all aspects of the character''s personality, because it doesn''t need to store individual aspects, only generalities.

I don''t know if that made any sense. But I think you get the point.

+------------------------------------------------
| Thp :: Threep2742@worldnet.att.net
| "Nature uses only the longest threads to
| weave her patterns, so each small piece
| of her fabric reveals the organization of
| the entire tapestry." -- Richard P. Feynman

The problem with this karma idea is too few variables. What you have described sounds fine for the purpose of your example but take it out of that and you have some bugs. For example, if you were taking into account two other groups, unrelated to politics, say for example two careers in the same variable - police officer and charity worker, the charity worker would have a much higher karma (we assume) and therefore the cop would get on better with the totalitarian dictator!
Just a quick example of why you need a variable for each attribute that you rate NPCs on, minimizing on these can have strange results when applied to a game.