Game AI Roundup Week #23 2008: 7 Stories, 3 Quotes, 1 Video

Novack on June 8, 2008

This week’s roundup is back to normal at, as we bring you links the web relating to game AI. This week was marked by impressions of Ubidays; also there have some solid articles and blog posts for you to read — as always. Remember, there’s also lots of great content to be found in the forums here! (All you have to do is introduce yourself.)

This post is brought to you mostly by Marcos Novacovsky (aka “Novack”) with minor editorial comments by Alex Champandard. If you have any news or tips for next week, be sure to email them in to editors at Remember there’s a mini-blog over at (RSS) with game AI news from the web as it happens.

Do You Find Me AImbitious?

Eric “SigmaX” Scott from the blog Thinking Aloud posted an article entitled “A Cybernetic Dream: Would a Tremulous A.I. be too Ambitious?“, about the process of thinking an AI system for the game Tremulous.

“I’ve been trying for a long time to form an intuition regarding just what is and is not reasonable to expect from simulated evolutionary processes. First my curiosity was drawn from a desire to determine whether abiogenesis is a plausible explanation for the origin of life [See ramblings here from Freshman year]. Upon my fairly recent discovery of artifical neural networks, this train of thought has been extended to find me wondering what all I can teach a computer to do.”

Solved AI

The blog Wikle Journal contains an interesting article entitled Artificial Intelligence = Solved on “how to solve the whole AI problem“.

“Artificial Intelligences is one of those problems that looks real simple till you try and solve it. Then it blossoms out to be stupidly difficult. [info]kvance once told me that the best idea any one had so far for a working AI was a series of searchable tables that would contain all human knowledge. The computer would then be able to act as an AI by compiling information from these tables. Of course creating a series of tables that contain all human knowledge is currently so difficult it needs its own word just to describe it. I’m going with ‘Malus Difficult’ cause it sounds latin.”

Ubidays 2008: Far Cry 2

The Ubisoft event left a lot of reviews. As a result, this week there where plenty of interesting comments on the Far Cry 2 AI. Sounds like many new exiting features, at least on its AI. Check out this comments:

QJ.NET: Far Cry 2 life-saving friendly AI

“Take for example, the intriguing implementation of friendly AI. Now first impressions of the AI system were, from early developer interviews at length, very much a revamp (if not an imitation) of GSC Game World’s A-life “needs” idea. But the artificial intelligence incorporates more than just needs and available resources.

Hands on demonstrations and a couple of gameplay screenshots showed that both the enemy and friendly AI will dash through firefights to gather their fallen and carry them out to safety where the casualties may be restored to fighting strength. The gameplay mechanic would include any character as long as they belong to a respective side - even you.

Say a plan of infiltration goes belly up and players find themselves suddenly seeing fleeting moments of their lives pass by. The visual queues don’t necessarily mean that the end is near.

Players will notice that while their mortality is in limbo, a companion they’ve met with earlier may rush to their aid, dash into a battle, fight off the buzzards (okay, kidding), and move players to safer areas.”

Kotaku: Making Far Cry 2’s Africa

“Next the team populated their artificial Africa with a blend of different wild animals, each with specific artificial intelligence. This animal AI tells the creatures when to sleep, hunt, even drink from a pool of water. They know to run from fire, and when to be started by approaching humans.

They tied all of this to the human enemy AI, allowing them to pick up on things like animal movement, so if you were to startle a creature into an enemy camp, it will alert the enemy.

I know this is starting to sound Fable-esque, but it they really can pull off all of this inter-species AI interaction it’s going to add a lot of realism to this game.”

play™: Far Cry 2 - preview

“AI is another area receiving a great deal of attention. The game is set in an open world, but throughout this you’ll make friends and enemies based on your decisions and actions, new paths opening depending on who you side with in a plot that revolves around the conflict between warring factions in a failed African state. Help one character, and he may help or advise you later, smite him and you might find he’ll come after you with an irate mob of hired ‘helpers’.”

Fracturing terrains

In an interview with Screen Play, Shara Miller (executive producer of the upcoming LucasArts third-person shooter Fracture) denoted the implications of new improvements to the context of the game, when cames to think and design the game AI.

“Q: How much of a technological challenge was terrain deformation?
A: It was a massive one. That’s the other thing that you’ll find about shooters. There might be objects in the world that you can move around, Havok physics are really big, it’s in everything, but you’ve never seen a game where the mesh, the actual ground, is movable. It creates major problems to be solved for AI (artificial intelligence) characters. What are AIs going to do in an environment that is constantly moving around? There is limited things that artificial intelligence can do, and if you throw anything in its path there’s going to be a host of issues, so that was a huge challenge. Just the memory that it takes to be constantly shifting the ground under your feet, all the calculations that have to be done for that, it’s a lot. And in multiplayer when you multiply it by 12 people doing that, it’s like, “Oh, wow!” You can’t really underestimate the challenge of a new mechanic like this on the tech side, it’s a pretty big deal.”

It’s All In The Wrist

Ted Vessenes (is the man is a machine?) wrote another good article on his blog BrainWorks about bot AI aiming realistically as a human player’s would do:

“Most games don’t let you spectate their AI units because it would be immediately obvious how the AI is cheating. But if you could, you almost always see on of two things. The AI might look almost directly at the player and but moves their camera away from the ideal position by a random amount, so that they miss sometimes at random. Or the AI could turn at a fixed, slowish speed, but always (eventually) aims more or less perfectly, so that they only miss right after the target suddenly changes.”

Final Version of AIFighter

We have been following the Billy Hewlett work on their final project for DESMA; it seems like the AI is finally complete.

“All game modes are working, the game is (almost) completely free of programmer art, and even includes an Easter Egg! Many playtesters really liked the ‘View AI’ option, but we never got the art working, so to access it the appropriate player should press the left trigger from the Player X Beats AI screen.”

WiC:Soviet Assault

In an interview with Skewed & Reviewed, Magnus Jansén (Lead Game Designer) talked about the World in Conflict, Soviet Assault AI.

“GVK: In games of this type, A.I. is often critical for the success of a game. How has the A.I. for the game been setup and what features will it include?

MJ: We built the AI from scratch to fit the team-play focus of World in Conflict. So when you are playing a skirmish game the different AI players will ask for assistance from other AI players depending on their need (although the player will not see this – it’s more how we laid things out). If you ask for their assistance they will respond. In the single player missions you often face units that are fully controlled by the level designers, so they don’t use the AI per se.”


Eurogamer featured a Live Q&A with Ted Price (CEO of Insomniac), about Resistance: Fall of Man 2. Here there are his words regarding an intersting feature on the game’s AI.

“For example, the traffic circle in Manchester was a real test of the efficiency of our AI. With so many enemies in a small area we found that the CPU was getting bogged down. So one of our programmers, Terry Cohen, wrote a great AI ‘LOD’ (level of detail) system which scaled the AI complexity based on enemies’ proximity to the player.”

Artificial Intelligence in Racing Games

This week we have the first video (Part 1 of 3) of a series about AI in Racing Games.

John Pile presents the history of AI in Racing Games during the 3 part presentation for Computer Games Technology at the University of Abertay in Dundee, Scotland

Stay tuned next week for more smart links from around the web!

Discussion 0 Comments

If you'd like to add a comment or question on this page, simply log-in to the site. You can create an account from the sign-up page if necessary... It takes less than a minute!