Accelerating Discovery with AI Surrogates: From Days of Simulation to Real-Time Insights

About the Session

Traditional scientific and engineering simulations—ranging from climate modeling to computational fluid dynamics—are highly accurate but computationally expensive, often taking hours or days to run. AI surrogates (also known as emulators or meta-models) are changing this paradigm. By training machine learning models on physical simulation or real-world observational data, researchers can approximate complex systems and get results in milliseconds instead of days.

This session provides a comprehensive, high-level guide to AI surrogates. We will clarify what surrogates are, distinguish them from digital twins, and explore how they integrate into the broader landscape of science, engineering, and business. Attendees will learn methodologies for creating surrogates from simulation and observational data, and gain a clear, decision-based framework on when to utilize them—and more importantly, when to avoid them.