Advancing Science and Security through Artificial Intelligence
LANL has successfully applied artificial intelligence (AI) technologies such as machine learning to multiple mission areas. Recognizing the accelerated breakthroughs in AI, especially in foundational models as demonstrated in natural language applications, LANL is making strategic investments to harness these potentially transformative approaches. For example, we have invested in Venado—a next generation computational platform optimized for AI.
Building on LANL capabilities and earlier investments, this SIC will ensure that LANL
- develops extensible, domain-specific, generative AI models in two challenge areas related to LANL’s mission by fine-tuning generic foundation models using high-quality domain-specific data and high-performance computing resources; and
- uses directed experiments and unique data to validate the approach and fill key knowledge gaps.
Potential challenge areas include nuclear security, biology, materials/manufacturing, and climate.
By investing in these domain-specific models, LANL will have a cutting-edge capability in AI that will then be available for use by the broader community to distill the relevant data available in many disparate unstructured forms into structured knowledge and to generate novel scientific hypotheses in the mission areas selected.3
This SIC directly supports the Critical Outcome computational & AI breakthroughs and has broad applicability across many of the Critical Outcomes in the Laboratory Agenda, including experimental advances; integrated deterrence; systems technology modernization; climate, clean energy, and national security; and biosecurity and emerging threats preparedness