Kodu AI Lab is a set of extensions to Kodu Game Lab, an introductory game making tool created by Microsoft. Kodu AI Lab, named because a number of the features are inspired by Artificial Intelligence techniques, works to expand the expressive potential of Kodu Game Lab by developing new elements that can support new game mechanics. The elements are meant to help children create games based on dynamic relationships, social interactions, and storytelling. The extensions are being developed and evaluated with middle school students. The first extensions (attitudes, learning, and fuzzy logic) are being guided by the principles of simplicity, understandability, and expressiveness. Simplicity means that the terminology we use is easy to grasp. Understandability means that the new language constructs will not result in unexpected behavior (a common challenge with many AI algorithms). Expressiveness means that the new features will allow users to intentionally create experiences that would have been difficult or impossible without the features. So far, we have developed three new language constructs for Kodu AI Lab. The first feature we call attitudes, which allows characters to put other characters into subjective classes and reference them as such. The second is learning, which gives characters the ability to dynamically classify other characters based on their behavior and other events in the world. The final feature is fuzzy logic, which allows characters to classify other characters to different degrees.
Click images to enlarge.
Teale Fristoe, Brandon Tearse, Eric Kaltman, Larry LeBron
Jill Denner (ETR Associates), Brad Gibson (formerly Microsoft Research), Matt MacLaurin (formerly Microsoft FUSE Labs), Michael Mateas, Noah Wardrip-Fruin
Expressive Intelligence Studio
Kodu at Microsoft