As AI research increasingly centers on building robust world models—algorithms that predict and reason about real-world dynamics—XR’s long history of creating richly simulated environments becomes indispensable. This session will explore how the XR community’s advances in photorealistic rendering, physics engines, and interactive virtual worlds provide the ideal playground for AI systems to learn, test, and refine their world models. Conversely, improved AI modeling techniques enhance virtual world fidelity, enabling ever more immersive, adaptive simulations.

We’ll compare goals and methods: world modeling seeks predictive accuracy (anticipating outcomes from inputs), while world simulation focuses on generating sensory- and behaviorally accurate environments. By uniting these disciplines, researchers can accelerate AI training with scalable, controllable data in virtual worlds, then validate those models in real-world scenarios. Join us to learn how this symbiosis fuels breakthroughs in robotics, autonomous agents, and next-generation XR experiences, and to discuss best practices for collaboration between AI and XR practitioners.