The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Deep Reinforcement Learning for Game Testing at EA with Konrad Tollmar - #517
Sep 9, 2021
In this discussion, Konrad Tollmar, research director at Electronic Arts and an associate professor at KTH, dives into the groundbreaking work of EA's SEED team. He reveals how deep reinforcement learning is transforming game testing, focusing on titles like Apex Legends and FIFA. Konrad highlights innovative methods for glitch detection using CNNs and the challenges of training AI in complex 3D environments. He also shares his vision for the future of AI in gaming, emphasizing the importance of adaptability for enhancing player experiences.
AI Snips
Chapters
Transcript
Episode notes
Half-Life for 3D Experiments
- Konrad Tollmar used Half-Life as a game engine for 3D experiments during his postdoc at MIT.
- This experience influenced his preference for game engines, even in later projects like self-driving car simulators.
Believable vs. Perfect AI
- Deep reinforcement learning (DRL) is promising but challenging to implement effectively in complex games.
- Game AI must prioritize believability over perfect performance, unlike other AI applications like self-driving cars.
Believable Group Behavior
- AI agents in games need to exhibit believable behavior, both individually and in groups.
- Their actions should reflect human-like tendencies, such as avoiding obstacles or collaborating towards a goal.