Artificial-Intelligence in Modern Video Games: Enhancing Realism and Player Experience

Abstract

The field of General Video Game Playing (GVGP) aims to develop autonomous agents capable of playing a wide array of video games without human intervention. In 2014, the General Video Game AI (GVGAI) competition framework was introduced to provide researchers with a unified, open-source platform to test AI algorithms on an extensive range of games created through the Video Game Description Language (VGDL). In pursuit of more immersive player experiences, we propose an innovative approach to modeling non-player character (NPC) behavior that extends beyond traditional emotion and personality-based methods. Our model incorporates cultural context to simulate NPC behaviors that reflect the values and norms of the character’s in-game region. Games, with their vast state spaces and inherent complexity, serve as valuable benchmarks for evaluating advanced techniques, including artificial intelligence (AI). Game AI encompasses methods that enhance game appeal, immersion, and intelligence. Given the limited recent reviews in this field, we present a systematic overview of key developments from 2018 onward, focusing on three core areas: believable agents in non-player character (NPC) research, procedural content generation for game level design, and player modeling through player profiling State-of-the-art fighting game AI technologies are already able to play at the level of professional human players