2025/09 - Practical Utility Theory for Game AI (David Rez Graham)
Formation créée le 30/06/2025.
Version du programme : 1
Programme de la formation
Use the decision-making techniques behind some of the most adaptable and intelligent game AI to create more believable characters. This course explores utility theory as a practical framework for AI behavior design, which is widely used in both indie and AAA titles. Taught by Rez Graham, an AI developer who has built a career around building systems for dynamic gameplay, you’ll learn how to apply these methods to evaluate competing goals, score actions, and create scalable, flexible AI for your own projects. September 22nd to 25th - 9am-11am 1. Introduction to Utility Theory & Implementation a. What is utility theory? b. Mathematics and Psychology c. Core components: scoring, curves, and goal evaluation d. Utility Architecture 2. Decision Design a. Variations on implementations b. Designing decisions c. Exercise: Designing an agent 3. Knowledge Representation & Analysis a. Internal knowledge representation and modifying behavior b. World analysis c. Hierarchical systems d. Introspection and tools 4. Real-World Analysis a. Case studies: examples from shipped games b. Deep analysis of real implementation c. Using utility theory outside of game AI d. Wrap-up, Q&A, and further learning paths
Objectifs de la formation
- Understand the core principles of utility theory and how they apply to AI decision- making in games.
- Learn to design scoring functions to evaluate different actions.
- Know how to combine an infinite number of considerations into a single decision.
- Master techniques for building scalable and flexible utility-based AI systems.
- Apply utility theory to non-AI systems to allow for more emergent gameplay.
Profil des bénéficiaires
- Gameplay Programmers
- AI Programmers
- Programming/Mathematics Background.
- A very basic familiarity with game AI.
- Masterclass in English
Contenu de la formation
-
Introduction to Utility Theory & Implementation
- What is utility theory?
- Mathematics and Psychology
- Core components: scoring, curves, and goal evaluation
- Utility Architecture
-
Decision Design
- Variations on implementations
- Designing decisions
- Exercise: Designing an agent
-
Knowledge Representation & Analysis
- Internal knowledge representation and modifying behavior
- World analysis
- Hierarchical systems
- Introspection and tools
-
Real-World Analysis
- Case studies: examples from shipped games
- Deep analysis of real implementation
- Using utility theory outside of game AI
- Wrap-up, Q&A, and further learning paths
David “Rez” Graham is a senior AI programmer currently working independently on a self-funded unannounced project. He is also the Director of Game Programming at the Academy of Art University’s School of Game Development where he built the entire Game Programming curriculum from scratch. He has worked in the games industry as an engineer since 2005 spending most of that time working on various kinds of AI, from platformer enemy AI to full simulation games. Most recently, he was the Lead AI Programmer on The Sims 4 at Maxis. Prior to that, he worked at PlayFirst, Slipgate Ironworks, Planet Moon Studios, and Super-Ego Games. He is the co-author of Game Coding Complete, 4th Edition, has two articles in Game AI Pro, one article in Game AI Pro 3, and regularly speaks at The Game Developers Conference. Rez spends his free time acting, performing magic, and running tabletop RPGs.
- Evaluation des compétences acquises via un questionnaire
Modalités de certification
- Délivrance d'un certificat de réalisation.