Artificial Intelligence for Computer Games

This editorial collection explores the current state of artificial intelligence (AI) in computer games, focusing on the transition from traditional rule-based systems to more advanced, adaptive, and autonomous methodologies. The primary thesis is that while commercial games have historically relied on simple scripts, integrating machine learning, multiagent systems, and evolutionary algorithms can significantly enhance gameplay, NPC behavior, and development efficiency. The document serves as a synthesis of nine peer-reviewed papers that examine these advancements across various genres, including real-time strategy, action, and mobile gaming.

The research highlights several key technical breakthroughs. Notable findings include the development of an abstraction-based moving target search algorithm that maintains performance in large problem spaces and a multiagent potential field architecture that rivals traditional bots in real-time strategy environments. Other contributions address the limitations of mobile hardware by utilizing lightweight machine learning for imperfect information games, and the application of genetic programming to automate the creation of terrain and the fine-tuning of emergent game environments. Additionally, the collection introduces novel frameworks for pathfinding aesthetics and the integration of multiagent systems into existing game engine architectures.

The scope of this research is global, reflecting academic efforts to bridge the gap between theoretical AI research and practical game development. The methodology across the included studies involves a mix of performance simulations, statistical analysis of pathfinding models, and comparative testing against existing game AI, such as finite state machines. By demonstrating that these advanced techniques can improve decision-making, predictive capabilities, and aesthetic quality, the research underscores the potential for more sophisticated, responsive, and autonomous gaming experiences.