Source metadata

  • Type: Academic textbook / field overview
  • Authors: Georgios N. Yannakakis and Julian Togelius
  • Published: 2018, Springer

Key takeaways

  • Defines the field broadly: AI in games includes playing games, generating content, and modelling players, not only controlling NPCs.
  • Introduces a unifying lens of representation and utility across almost all AI methods used in games.
  • Treats behaviour trees, utility AI, tree search, evolutionary computation, supervised learning, reinforcement learning, and unsupervised learning as one connected methodological landscape.
  • Gives high-value overview material on procedural content generation taxonomy and roles, especially mixed-initiative, autonomous, experience-driven, and experience-agnostic generation.
  • Gives one of the clearest broad treatments of player modelling in games: what it is, why to do it, what data it consumes, and what outputs it can produce.

Notable claims

  • The book explicitly argues that games are useful for AI, and AI is useful for games; the relationship is two-way.
  • It treats player modelling as a central game-AI concern because modelling behaviour and experience enables adaptation, analysis, and personalisation.
  • It highlights ethics and general game AI as frontier concerns, not optional add-ons.

Relevance

Directly informs:

Supports:

Open questions raised

  • How much of the AI-and-Games research view should be surfaced in lecturer-facing synthesis pages versus student-facing implementation pages?
  • Which parts of the player-modelling chapter should be connected more tightly to the existing analytics and psychology sections?

overview-artificial-intelligence-in-games · game-ai-agent-design · machine-learning-games · procedural-generation · player-modelling · behaviour-trees · neuroevolution · utility-ai · mcts