Behaviors of Actors in a Resource-Exchange Model of Geopolitics

Abstract

We present initial findings of an ongoing effort to endow the key players in a nation-state model with intelligent behaviors. The model is based on resource exchange as the fundamental interaction between agents. In initial versions, model agents were severely limited in their ability to respond and adapt to changes in their environment. By modeling agents with a broader range of capabilities, we can potentially evaluate policies more robustly. To this end, we have developed a hierarchical behavioral module, based on an extension of the proven ATLANTIS architecture, in order to provide flexible decision-making algorithms to agents. A Three-Layer Architecture for Navigating Through Intricate Situations (ATLANTIS) was originally conceived for autonomous robot navigation at NASA’s JPL. It describes a multi-level approach to artificial intelligence. We demonstrate the suitability of our reification for guiding vastly different types of decisions in our simulations over a broad range of time scales.