Today is just one of those days I’d rather forget… Hmph. If you have a few seconds to spare and a StumbleUpon account, could you post a short review of AiGameDev.com? It’d certainly help put the spring back in my step. Thanks!
Anyway, this Saturday, there are quite a few interesting Smart Links about game development and artificial intelligence! Feel free to contact me if you have any news or tips for next week. Also subscribe to the mini-blog for news about game AI as it happens…
References in Intelligent Narrative Technologies
Andrew Stern over at Grand Text Auto has posted an extensive summary of the Intelligent Narrative Technologies 2007 symposium. He drops lots of great references that’ll take you weeks to read up on! It’s like candy for game AI developers :-)
Intelligent Narrative Technologies 2007, grandtextauto.org
Evolving Neural Networks in Polyworld
There was a Google Talk last week about using evolutionary algorithms to turn CPU cycles into brain designs. This is done in a simulation called Polyworld, in which agents are created as free-form neural networks. Polyworld is the brainchild of Larry Yaeger, but the talk is by first year Caltech graduate Virgil Griffith.
This is a preview only. Click below for the full video.
As you can see, the results are far from reaching the singularity (as suggested by the video’s summary), but the question is: are we getting any further with this approach?
Reactions to the Polyworld Video
While applying such ideas in Polyworld may be fun and interesting research, many feel that very little progress has been made using GA to evolve NN behaviors in such simulated worlds. This discussion on the Artificial General Intelligence mailing list sums up what other researchers are thinking:
“Although I thought this was a good talk, to me it seems fairly clear that little or no progress has been made in this area over the last decade or so. In the early 1990s I wrote somewhat similar simulations where agents had their own neural networks whose architecture was specified by a genetic algorithm, but just like the speaker I came up against similar problems.” — Bob Mottram
So it seems GA and NN are having just as hard a time as any other AI technology at reaching general intelligence — if not harder. Do you consider this kind of research potentially more fruitful?
[AGI] Polyworld: Using Evolution to Design Artificial Intelligence, mail-archive.com
An Overview of the Artificial Intelligence Field
Develop magazine makes a brief tour of the game AI industry, interviewing developers from various middleware solutions, including PathEngine, Kynogon and Xaitment. Here’s the hook from the introduction:
“There’s a bit of a problem with AI. But it’s not some insurmountable scientific barrier, and neither is it a technological constraint. Rather, the problem is that it largely exists in the eye of the beholder.”
The article is writted by Ed Fear.
An Overview of the Artificial Intelligence Field, developmag.com
Pyke: Knowledge-based Inference Engine
Pyke has released a new version. It’s a knowledge-based inference engine (expert system) written in 100% python that can do both forward-chaining (data driven) and backward-chaining (goal directed) inferencing. It also automatically generates python programs by assembling individual python functions into complete call graphs.
Pyke: Knowledge-based Inference Engine, sourceforge.net
Cinematic Video Games
Peter Plantec over at VfxWorld digs deep into EA’s plan to push the boundaries of games and make them more immersive and emotionally engaging like movies. Here’s the only interesting quote about game AI:
“Both LaBounta and Zargarpour [from Electronic Arts] discussed how complex the process of developing cutting edge games is. They talked about the enormous number of possible scenarios that need to be created and about improving game AI so that the game itself is more aware of what the player is doing and how to respond most effectively. Clearly, from my point of view, EA wants to re-stimulate burned out game players who are getting tired of sequel after sequel”
Do you think AI has potential for generating such personalized content?
Can Videogames Be Made Truly Cinematic?, vfxworld.com
Some AI Engine Design Considerations
Jose Enrique explains general considerations for building game AI. He gives an overview of how the entire system should look like once finished and continue to explain each particular piece in detail. He provides an overview of a variety of techniques to go “beyond the state machine.”
Story + Games + Math = Call for Papers
The Symposium on Logic and the Simulation of Interaction and Reasoning in Scotland is requesting submissions:
“Have something interesting to say on game design, interactive storytelling, characters, and their corresponding methods of simulation? During this symposium mathematicians, computer scientists, and game designers will collaborate to chart interesting paths in the realm of mathematical approaches to storytelling and interaction design in games.”
This is certainly an up-and-coming field. Let us know if you’re thinking of participating!
Story + Games + Math = Call for Papers , usc.edu
On the Difficulty of Integrating AI and Animation
Next Generation has an interesting post-mortem about Call of Duty 4. There’s an interesting quote about artificial intelligence in there:
“The point where AI and animation meet is a difficult thing to get right. Ideally, you need to find people with programming ability, an eye for animation and a good sense of gameplay. Alternatively, you can pair people with programming ability and an open mind with experienced animators and gameplay designers.
Either way, you have to remember that this is not an easy task. However, if you’re putting significant effort into making great animation, you are wasting that effort if the animation doesn’t play well with your AI.”
If anything, this goes to emphasize the strength of multi-disciplinary teams, as inspired by Scrum. What do you think?
Post Mortem: Call of Duty 4, next-gen.biz
Adaptive Player Modeling Using AI in Pacman
ZenBen is running an experiment as part of his Ph.D. project entitled Player Profiling & Modelling for Adaptive Artificial Intelligence in Computer and Video Games. His experiment aims to address the thorny problem of building into a game, an A.I. that can reason about the player’s preferences for the kind of experience they will have.
More details about this tomorrow, but feel free to help out now by downloading the game and playing it.
Adaptive Player Modeling Using AI in Pacman, zenben.net
Artificial Intelligence and Computer Gaming
Over at Demerzel’s blog, they ponder about the benefits of an AI that does not cheat: “Why is it so difficult for game designers to essentially separate the game AI from the bot AI such that the bots have the same disadvantage you do? Essentially strip away its ability to see/know everything about your resources and let it at least feel a little more human”.
Artificial Intelligence and Computer Gaming, micahfk.com
Stay tuned next Saturday for more Smart Links from around the web.