Frontier AI for Science, Security, and Technology (FASST): transformative AI for the fusion community and the nation*

AI is an incredibly promising tool to help our fusion community integrate information from an ever-expanding range of experimental facilities, diagnostics, target fabrication and metrology sources, as well as new theory and design practices. To fully realize the potential of AI methods for the US in fusion and other applications, we require a large-scale, transformative effort aimed at rapidly advancing scientific AI capabilities. The Department of Energy (DOE) has built such an initiative, called FASST: Frontier AI for Science, Security, and Technology. FASST will curate massive scientific data resources, augment world-leading computing capabilities for AI, and build new frontier-scale scientific AI models and systems. In this talk, we will describe the goals and directions of FASST AI research. We will highlight key connections to fusion systems with an emphasis on digital twins. This will include descriptions of AI-driven autonomous labs and production facilities – physical and digital twins that can join science in the lab with science in the computational data center to produce insight at unprecedented rates. We will cover specific examples from fusion platforms, high-repetition rate HED laser systems, advanced manufacturing target production, and more. We will end with a view of the key research directions and opportunities provided to the fusion community by the coupling of FASST AI methods and digital twins.