Foundation models together with learned game‑world models are reshaping AI throughout the entire game lifecycle. Beyond classic game playing, recent systems also model player behavior and game dynamics, assist design and development, adapt player‑facing experiences at runtime, and evaluate the resulting artifacts. The literature can be organized into six roles: playing and acting, modeling players and games, designing games, building and maintaining games, generating and adapting at runtime, and testing and evaluating games.
For each role we consider the structure supplied by the game or workflow, what the AI learns or produces, which capabilities and artifacts transfer across settings, and the evidence supporting the claims. Cross‑role connections include: trajectories train world models, learned environments provide experience for agents, design specifications drive executable implementations, and play or testing feedback guides revision.
Nevertheless, control schemes, rules, engine interfaces, state representations, and player contexts often remain setting‑specific, so downstream claims must be validated in the target environment. Evaluation is relatively standardized for bounded game playing and selected learned environments, while persistent state in learned worlds, repeated software revision, validated player modeling, sustained runtime adaptation, and representative automated testing are less established.
The central challenge is to reuse or transfer AI outputs and capabilities across roles while re‑establishing evidence of effectiveness in each game‑specific context.
Review