Intelligent agent modeling as serious game

Abstract

We propose a novel approach to intelligent tutoring gaming simulations designed for both educational and inquiry purposes in complex multi-actor systems such as infrastructures or markets. Rather than letting students perform guided role-play, the idea is to guide students as they construct, learn about and exchange delegate agents to test operations strategies in various simulated scenarios. While various technologies for end-user agent modeling and intelligent tutoring for ill-defined domains already exist, they are yet to be combined into a single intelligent simulation platform. We propose users construct the behaviors for their delegate agents at their own level of expertise. That is, simply selecting and adjusting some predefined agent behavior, shaping the desired behavior as it evolves in user-defined train scenarios, enacting example behavior for the agents to imitate, or building the behavior model in detail with simplified programming languages. The envisioned platform is a low-threshold interface for knowledge exchange and discovery where learning takes place between students, from agent to student and vice versa.