Game AI Roundup Week #16 2008: 5 Stories, 1 Video, 1 Paper, 2 Quotes, 1 Book

Novack on April 20, 2008

Weekends at are dedicated to rounding up smart links from the web. This week you’ll find a few great blog posts, articles, another Google Talk, and a new white paper. 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 by Novack and Alex Champandard. If you have any news or tips for next week, be sure to email them in to editors at Also Remember there’s a mini-blog over at (RSS) with game AI news from the web as it happens.

Real AI for Games

Dave Mark’s latest article Why Not More Simulation in Game AI? here on, elicited an interesting debate in the comments. Paul Haley in particular, posted on in-depth reply on his blog Commercial Intelligence.

“Dave Mark’s post and the comments it elicited are right on the money about the correlation between lifespan and intelligence of supposedly intelligent adversaries in first person shooter (FPS) games. It is extremely refreshing to hear advanced gamers agreeing that more intelligent, longer-lived characters would keep a game more interesting and engaging than current FPS. This is exactly consistent with my experience with one of my employers who delivers intelligent agents for the military. The military calls them “computer generated forces” (CGFs). The idea is that these things need to be smart and human enough to constitute a meaningful adversary for training purposes (i.e., “serious games”).”

Enemy Territory AI

Building intelligent multiplayer bots that can compete against hardcore gamers may seem like the Holy Grail of game AI. WorthPlaying published a preview of Enemy Territory: Quake Wars, the console (Xbox 360 and PS3) version, based on PC game developed by Splash Damage. They commented about AI:

“With “asymmetric gameplay,” the characters of both the GDF and the Strogg look, move, and behave uniquely, while advanced Artificial Intelligence (AI) technology provides formidable friends and foes in online and offline play, allowing new players to ease into the game, while intuitively challenging more advanced ones.”

Also be sure to catch up on some previous interviews with the original creator of the PC bots, John Dean.

Getting it “Just Right”

Another great post from Ted Vessenes on his blog BrainWorks, focusing on AI numbers and measurements. Here is an extract of the best paragraphs:

“When you design artificial intelligence, everything needs to be broken down to numbers, since a computer can only understand numbers. And a lot of times you have no idea what numbers best simulate the behavior you want to elicit from the AI, or even if the number encodes the correct concept. […]

All this to say that when you write AI, there’s an awful lot of number tweaking going on. Either you do it by simulations, by neural network training, genetic algorithms, or by hand. But one way or another, you’ll have dozens of numbers that all need to be “just right” to present the illusion of intelligence. The good news is that once you’ve spend time finding these values, you’ll have excellent insight into how humans approach the problem. That’s how research works, I guess.”

The Next Generation of Neural Networks

Over in the increasingly popular and ever useful forums, Phil Carlisle pointed out another fascinating Google Tech Talk. The speaker is Geoffrey Hinton, a prominent figure in the field on connectionism over the past decades.

“In the 1980’s, new learning algorithms for neural networks promised to solve difficult classification tasks, like speech or object recognition, by learning many layers of non-linear features. The results were disappointing for two reasons: There was never enough labeled data to learn millions of complicated features and the learning was much too slow in deep neural networks with many layers of features. These problems can now be overcome by learning one layer of features at a time and by changing the goal of learning. Instead of trying to predict the labels, the learning algorithm tries to create a generative model that produces data which looks just like the unlabeled training data. These new neural networks outperform other machine learning methods when labeled data is scarce but unlabeled data is plentiful. An application to very fast document retrieval will be described.”

It seems like the decades of mathematical research are finally paying off!

A Paper about Realistic AI and RTS Games

As partial fulfillment of the requirements for completion of a Bachelor of Technology Program, the Dane Anderson wrote this paper and published it on Scribd. (Dane is another active member of the community & forums here, so if you check the references section you shouldn’t be surprised to find among the sources.)

“The purpose of this paper is to answer the central question, “Can a Realistic Artificial Intelligence be created for a Real-Time Strategy Game?” A brief framework on how to achieve a realistic real-time strategy (RTS) game artificial intelligence (AI) is herein outlined in order to demonstrate how RTS games enable the AI to make optimal decisions in real-time.”

Can a Realistic AI be Created for a RTS Game?
Dane Anderson
Download PDF

Artificial Stupidity

GamesRadar posted am article on AI quirks, based on a compilation of youtube videos. Strange coincidence, as here on AIGameDev there is a comming -technical- article on the same direction.

Having played hundreds of games over the years, we have to say true artificial intelligence is rare. Does Seaman really learn to speak? Did the AI in Forza 2 really learn how to double-brake? Who knows. What we do know, however, is that it’s very easy to confuse a computer – sometimes without doing anything at all.

We can happily type that into one without fear of reprisal as the office PCs don’t have robotic arms (yet). So for now, we can laugh – as will you.

New Book: Game Artificial Intelligence

Thomson/Delmar Learning just published a new book on the “Game Development Essentials” series. Writen by John Ahlquist and Jeannie Novak, the book is entitled Game Artificial Intelligence. Here is a brief content description:

“History and concepts : how did we get here (and where is here anyway?) — AI agents : creating behavior using a finite state machine — Complexity & broad scope : multiple states and hierarchical state machines — Customizing AI systems : allowing others to modify and implement the AI — Expert systems : capturing high-level knowledge and improving behavioral control — Pathfinding : allowing agents to properly find and follow paths — Advanced pathfinding techniques : improving performance and quality — Looking smart : maximizing perception of intelligence — Putting it all together : designing and building quality game artificial intelligence”

It’s currently at 32% discount on Anyone with a review of the book to share?

New game with xaitment’s AI

A short press release about xaitment’s AI being used on “Alarm for Cobra 11 - Crash Time” Game:

“xaitment GmbH, one of the leading developers and service providers of artificial intelligence for the games and simulation industries, announced today that Synetic - The Funfactory, a German game developer specializing in the creation of racing games, used xaitment’s artificial intelligence on their latest title, the Xbox®360 version of “Alarm for Cobra 11 - Crash Time”. The game, which is being published by RTL Games, is scheduled to release this Spring.”

AI and General Design

Brian Edey, from the Canadian gaming website Game Focus, wrote a review on Universe at War: Earth Assault the Petroglyph RTS, published by Sega. He described an interesting problem, that clearly shows the delicate composition of elements in a game; in this case an AI controlled enemy could result overwhelming, due to design desitions not necesarily related with the AI itself.

“You can only build six units in que at any time for some strange reason in a game from 2007 we see this when every other game allows you to put in an order for a huge number of units. The developers must not have realized the AI auto builds and the player has to build manually and this shoves the balance of the game versus the AI all out of wack.”

Call for Submissions: Journal of Gaming and Virtual Worlds

For those of you looking for another outlet for your research ideas, here’s another option:

“The Journal of Gaming and Virtual Worlds is a peer-refereed, international journal which focuses on theoretical and applied, empirical, critical, rhetorical, creative, economic and professional approaches to the study of electronic games across platforms and genres as well as ludic and serious online environments such as Massively Multiplayer Online Role Playing Games and Second Life. The Journal aims at researchers and professionals working in and researching creative new media and entertainment software around the globe and seeks to document, harmonise, juxtapose and critically evaluate cutting-edge market trends, technological developments, as well as socio-cultural, political, economic and psychological concerns. It informs its readers about recent events such as conferences, and features long articles, short papers, poster abstracts, interviews, reports and reviews of relevant new publications, websites, virtual environments and electronic artefacts.

Contributions are invited from all fields of game studies research, design and development. We seek to provide a platform for vivid information interchange between academia and industry, between scholarship and professionalism, between theory, criticism and practice”

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!