Games for Artificial Intelligence Research: A Review and Perspectives

TL;DR

The paper surveys publicly available games and game-based platforms as AI test-beds, taxonomy, and guidance for playing and designing with AI. It methodically catalogs platforms by genre, purpose, agents, observability, languages, and competitions, and analyzes the challenges new game environments pose for AI, including high dimensionality, exploration, partial observability, generalization, and multi-agent dynamics. It contributes a practical roadmap for researchers and educators, detailing how to select appropriate environments for learning-based or planning-focused AI, as well as how to leverage AI for game design, tuning, and content generation (including PCG and LLM-assisted creation). By examining open-world and open-ended environments alongside emergent LLM capabilities, the work highlights opportunities for AI creativity, interdisciplinarity, and potential real-world impact, while calling for more accessible, well-documented platforms and broader genre coverage to accelerate progress toward AI that can both play and design games. The synthesis underscores the bidirectional influence between game AI and AI research, with implications for AGI, safety, and human-AI collaboration in creative contexts.

Abstract

Games have been the perfect test-beds for artificial intelligence research for the characteristics that widely exist in real-world scenarios. Learning and optimisation, decision making in dynamic and uncertain environments, game theory, planning and scheduling, design and education are common research areas shared between games and real-world problems. Numerous open-source games or game-based environments have been implemented for studying artificial intelligence. In addition to single- or multi-player, collaborative or adversarial games, there has also been growing interest in implementing platforms for creative design in recent years. Those platforms provide ideal benchmarks for exploring and comparing artificial intelligence ideas and techniques. This paper reviews the games and game-based platforms for artificial intelligence research, provides guidance on matching particular types of artificial intelligence with suitable games for testing and matching particular needs in games with suitable artificial intelligence techniques, discusses the research trend induced by the evolution of those games and platforms, and gives an outlook.

Figures (6)

• Figure 1: Roadmap of games and platforms, created by authors using screenshots of the games and platforms and figures of papers cited in this work.
• Figure 2: Screenshot of OpenAI Five berner2019dota.
• Figure 3: Games and platforms for playing, categorised as board/card, grid-based 2D, pixel-based 2D, 3D, social and text game types.
• Figure 4: Summary of games and platforms used for studing AI for game design, created by authors using screenshots of the games and platforms, as well as figures from their corresponding papers, reviewed in Section \ref{['sec:design']}.
• Figure 5: Mario level segments generated by the work of wang2024negatively tested and visualised in Mario AI framework karakovskiy2012mario.