Archived

This topic is now archived and is closed to further replies.

techpage

What about Academic AI and Game AI?

Recommended Posts

mjacobsca    122
I think this topic was discussed elsewhere on this forum, so you might want to check it out.

But "Game AI" typically refers to the elements that make a computer opponent play automatically - pathfinding, unit building, resource management, etc... Game AI can include cheating and other hard coded scripted events. It doesn''t matter if the AI can learn, as long as it can win the game against the human player.

"Adademic AI" typically refers to the goal of creating an AI that can learn, or respond to events though it''s algorithms. This level of AI has a much more lofty goal than in most game engines.

Share this post


Link to post
Share on other sites
techpage    122
thanks Gaiiden, the page you mentioned is full of Game AI vs. Academic AI. I''ve also found it in my Game Programming Gems II book in its introduction section by steven woodcock. He wrote Academic AI emphasised on the PROCESS and Game AI on the RESULT. He also gave the example with regards to pathfinding. While Academic AI avoids inadmissible paths, Game AI takes full advantage on them to speed the search.

Well... do you have other examples? like for example, in strategy games such as C&C?

Share this post


Link to post
Share on other sites
alexjc    457
It''s all about requirements. Academic AI sets itself well defined problems, whereas for game AI deadlines are more important. Hence the reason why scripts often aren''t sufficient in academic AI... try getting a Ph.D. with a script interpreter (lot of work, not interesting).

As game AI becomes more complex, academic AI algorithms will shine through... (B&W is the first example, with many other to follow

I have a lot of respect for Steven, but his intro to GPG2 is misleading and inaccurate.



Artificial Intelligence Depot - Maybe it''s not all about graphics...

Share this post


Link to post
Share on other sites
techpage    122
quote:
Original post by alexjc
I have a lot of respect for Steven, but his intro to GPG2 is misleading and inaccurate.



Artificial Intelligence Depot - Maybe it''s not all about graphics...


No bad intention, but which part do you think is misleading and inaccurate? Umm... I''m gonna use the book as part of information for my thesis...

Share this post


Link to post
Share on other sites
alexjc    457
techpage, two bits in particular.

1) Inaccurate - That academic AI is snobish with regards to the algorithms chosen... That''s usually due to tougher requirements for the problem at hand, where hard-coded behaviours won''t fit... games are simple compared to some of the problems tackled.

2) Misleading - That academic AI does not focus on the resulting behaviour. True, you have to come up with new material to get funding... but the reasons for modifying the process is about getting better behaviour in a simpler fashion.



Artificial Intelligence Depot - Maybe it''s not all about graphics...

Share this post


Link to post
Share on other sites