quote: Original post by Geta
Ok, so how do we do that?

Hey, I never said I had ''The Answer'' (TM)!!!

Although I do believe that ''an answer'' would have something to do with opponent modelling. It''s something we''ve been working on here at Monash for playing Poker. It is possible to build an online model of each opponent based on their activities in the game. This model can then be used in planning and decision making while playing the game... and I would go so far as to suggest that it could be used to make the game harder or easier for different players, based on their perceived skill levels (although this would definitely depend on the genre).

This kind of modelling is still in its infancy, so I wouldn''t expect to see it in games for quite some time... but I do think it is a very important step for the future.

Cheers,

Timkin

Opponent modelling is already being used in the AI of FPS bots Tim. I think William van der Sterren was the first to introduce it (to FPS bots that is)in his halflife bots.





Stimulate

No I''ve not seen any modelling for those types of games either Eric. I imagine it''s a lot harder to do though.

Thanks for that link. I''ve not seen that article before. Do you gave any others of interest?



Stimulate

I want to qualify my comments here first by saying that I''m not an AI programmer, and have only ever done theoretical research on the topic. Still, it''s one that interests me greatly.

I think the problem with trying to make AI ''Humanlike'' is that what defines ''Human'' behaviour can vary greatly. Do we want to emulate childish behaviour that currently rules online games? Probably not - then again, that''s almost certainly not what was being suggested. But playing against a computer AI who ''acts'' as a computer is not enjoyable either - I''m remembering one RTS in particular where the AI always saw the whole map and never forgot your locations. Using ''superweapons'' he could target specific buildings without ever exploring your section of the map; worse, even after building a stealth generator, he could still target those buildings (Although it did seem to reduce the frequency of the attacks). Personally I detest AI that uses ''special'' abilities to make up for it''s weakness.

What if the computer had to learn the map though, and could only ''see'' what a similar human player could? Would that be acceptable? In this same game, a human player could also target unexplored (or shrouded) areas of the map too - but you never knew what you might hit. So in the case of a stealth generator, if you used a unit to explore and uncovered the building, you could target it based on it''s remembered location -- that is, if you could remember exactly where it was. So perhaps you would actually want to model forgetfulness into the AI - after all, if it could remember perfectly the location of the buildings it uncovered, would it really be that different from knowing exactly where they were to begin with?? And if you did, wouldn''t that be giving the computer AI ''Human'' traits?

In short, I think you have to specify what you''re trying to incorporate, be it Adaptability (another Human trait), Learning, Deviousness, etc.

Just my two cents.....

quote: Original post by fup
Opponent modelling is already being used in the AI of FPS bots Tim. I think William van der Sterren was the first to introduce it (to FPS bots that is)in his halflife bots.

Some additional FPS ''bot AI opponent modeling can be found in:

It knows what you are going to do: Adding Anticipation to Quake Bots

I have yet to encounter anything for turn-based or real-time strategy games.

Eric

quote: Original post by fup
No I''ve not seen any modelling for those types of games either Eric. I imagine it''s a lot harder to do though.

I agree! Unfortunately (or fortunately depending on one''s perspective), those types of
games are the games I seem to end up getting contracted most to develop the AI for.

quote: Original post by fup
Thanks for that link. I''ve not seen that article before. Do you gave any others of interest?

I have thousands of links! I keep them all in google!

Seriously, if you are interested in ''bot AI, it is worth reading Dr. John Laird''s papers.
He is a super nice guy who is quite an advocate of using commercial computer games
in academic AI research. http://ai.eecs.umich.edu/people/laird/index.html

Eric

Why do you start off assuming that a soul exists? It''s just silly.

You may say there is no evidence that a soul does not exist by why does that therefore mean it might? There is no evidence the universe is inside a goats stomach but that doesn''t mean I should consider it true.

Does a beetle have a soul? They''re not too bright.
Does a squirell have a soul? They have quite good problem solving abilities.
Does a baby have a soul? They are too young to solve problems.
What about a really dumb adult?

At what point in life does something have a soul? What does it do? Why would one living thing have a soul and one not.

What evidence at all is there that a soul exists? After all the time that scientists have spent studying the human body, the universe, calculs, NN and the fact that you want to model human intelligence, I am so surprised that people are still clutching onto fairy tales.

quote: Original post by Geta
I have thousands of links! I keep them all in google!

Don''t be cheeky ;0). I must have done hundreds of searches for bot AI but, although I''ve heard of SOAR, I''ve never seen that article. Was looking around the SOAR website just last week actually and never spotted it. Can''t see the wood for the trees sometimes... especially after a week of 15 hour days.







Stimulate

quote: Original post by Anonymous Poster

What evidence at all is there that a soul exists? After all the time that scientists have spent studying the human body, the universe, calculs, NN and the fact that you want to model human intelligence, I am so surprised that people are still clutching onto fairy tales.

Get a grip. People have a right to their beliefs, no matter what they are. Further, I still haven''t seen anyone in this thread dogmatically say there was a soul... merely taking the possiblity into account.

Webster''s Online:
One entry found for narrow-minded.


Main Entry: nar·row-mind·ed
Pronunciation: "nar-O-''mIn-d&d, "nar-&-
Function: adjective
Date: 1625
: lacking in tolerance or breadth of vision : PETTY
- nar·row-mind·ed·ly adverb
- nar·row-mind·ed·ness noun

''nuff said

quote: Original post by iQ
Get a grip. People have a right to their beliefs, no matter what they are. Further, I still haven''t seen anyone in this thread dogmatically say there was a soul... merely taking the possiblity into account.

Webster''s Online:
One entry found for narrow-minded.

Main Entry: nar·row-mind·ed
Pronunciation: "nar-O-''mIn-d&d, "nar-&-
Function: adjective
Date: 1625
: lacking in tolerance or breadth of vision : PETTY
- nar·row-mind·ed·ly adverb
- nar·row-mind·ed·ness noun

''nuff said

Posting a definition to a word does not prove anything. I could post a dictionary definition on what being right means but that doesn''t prove that I am.

Do you believe the universe is inside a goat''s stomach? Does it make you narrowminded if you don''t? No, it doesn''t, that is just silly. Just because I don''t agree with something does not make me narrowminded.

Please enlighten me; what is a soul? What does it do? How are you aware that one might exist? Do insects and animals have soul? If they don''t, why not? (Not rhetorical questions)

As far as I know, a soul is something that you have that makes you human in a feelings way. Animals do not have soul. Animals can learn, we can learn, therefore even if a soul does exist (which I think is silly), it does not effect thinking.

I just find it very annoying that all the work that scientists do to prove things counts for nothing because of issues like this. Why do we have to humor people with ideas that have nothing to do with fact? Why can''t they be wrong until proven true like everything else in life?

I think human learning comes about because trees watch us do things and they tell our brain telepathically what we are doing wrong so we can change to get the right results. There. Why don''t you take that possibility into account? It could be true also.