Intelligent Agent Modeling as Serious Game

Educators increasingly turn to serious games to let students explore complex worlds in a safe environment. In serious games for ill-defined problem domains such as infrastructures and markets, students often interact with preconceived agents at an operational level. We hypothesize however that students could discover more about a domain’s complexity at a strategic level by building and testing their own delegate agents. Testing this requires an environment where students and teachers can construct agents at their own level of expertise with recent modeling technologies. For instance, students may create agents not just directly, by building or modifying comprehensive agent models with visual programming languages, but also indirectly, by shaping agent behavior as it evolves in user-defined training scenarios or by enacting example behavior which agents learn to imitate. We propose a serious game concept that combines such modeling methods within a single intelligent simulation platform so that it becomes a low-threshold interface for continuous knowledge exchange and gain between teachers, students and agents.