Abstract

Genesis Mission is a Department of Energy AI effort aimed at recruiting cutting-edge AI for science in order to double US R&D productivity while delivering innovation overmatch: the ability to mitigate emerging threats on timescales faster than those of U.S. adversaries. The effort is organized around two coupled thrusts. The first is a set of Science and Technology Challenges that define high-value problems across science, national security, and applied energy, thereby providing the mission pull for development. The second is the Genesis Platform, a federated, agent-first platform that connects frontier AI models, trusted data, high-performance computing, laboratory facilities, and mission workflows through governed and reusable interfaces.
This talk will describe the top-level goals and architecture of Genesis and explain how the Platform is intended to transform isolated demonstrations into durable operational capability. I will present examples and demonstrations spanning national security, applied energy, and scientific discovery to illustrate how AI-enabled workflows can support planning, analysis, orchestration, and decision-making across heterogeneous computational and experimental environments.
I will also discuss why Genesis is relevant to the applied mathematics community. Core technical challenges include modeling and simulation, inverse problems, optimization and control, uncertainty quantification, scientific machine learning, surrogate modeling, verification and validation, and reliable autonomous workflow execution. These areas place applied mathematics at the center of both the scientific opportunity and the rigor required for mission deployment. The talk will conclude with a call for engagement from the applied mathematics community in methods, benchmarks, evaluation, and co-design.