Beyond Equilibrium: Utilizing AI Agents in Video Game Economy Balancing

Abstract

In the era of "Games-as-a-Service,"a large number of game studios face the challenge of creating immersive virtual economies that offer enjoyable experiences for players. Evaluating game economies and sustaining their balance can be a complex and costly task, which often relies on human play-testers. To address this, our work explores the utilization of simulated AI players as a viable solution. This Work in Progress article presents our preliminary analysis of fundamental entities that construct game economies and the variables of these entities that impact different aspects of player experience. The focus is on five basic economic entities: Currency Sources, Inventories, Converters, Currency Sinks, and Traders. By examining these entities and their defining factors, we determine their influence on player experience. Subsequently, we propose AI Agents that are capable of playing games with a variety of play styles and adjusting the mentioned variables, ultimately enhancing the player experience based on the developers' goals. This approach offers game developers a cost-effective means of evaluating and optimizing in-game economies, without having to spend excessive time and money on human testers. Our research contributes to the development of coherent virtual economies by providing fast and accurate feedback, reducing costs, and making economy balancing more accessible to indie developers.