Artificial intelligence and machine learning are emerging as important tools for addressing LLNL national security and science missions. We’re applying AI/ML and cognitive simulation everywhere to increase the predictive power of broad range of simulation models, design new molecules for needs ranging from pandemic response to energetic materials, integrate complex multimodal data for national security decision support, enhance the next generation of predictive climate models, and much more. With a wide range of opportunities and available resources, the Lab is developing a deliberate and focused vision for AI development and execution.
The goal of the AI Innovation Incubator (AI3) is to advance AI for applied science at scale. The AI3 expands LLNL's capabilities through industry collaborations, establishes visible leadership in AI for applied science, develops informed strategies for mission-driven AI investments, and coordinates investments focused on exploring and developing AI.
How Does the AI3 Work?
Pillar 1: Supporting Community Vision
The AI3 is designed to organize and communicate a common view of AI for applied science, which is a multi-pronged effort. A first crucial step is to establish and communicate a coherent vision and framework for AI-focused activities, including publishing commentaries and reviews for a public audience as well as hosting collaborative project-sharing meetings and workshops. This effort also includes informing and guiding priorities and policies of sponsors and national stakeholders. Our goal of building a community of practice around AI for applied science relies on outreach to make connections and nurture win–win relationships.
Pillar 2: Collaboration Hub
Multidisciplinary collaboration is a key component of the AI3. Working with experts across many domains helps connect our R&D to external stakeholders, opening the door to new opportunities. Datasets from our unique scientific and experimental facilities can be shared with collaborators. We can also provide benchmarks for scientific AI algorithms, methods, and workflows for a range of scenarios: physics-based learning for biology, climate, and fusion energy; molecular generative models for materials and drug discovery; fast, flexible learning workflows for advanced manufacturing and self-driving experimental facilities; and data and computing ecosystems for large-scale scientific AI. Read more about our collaborators below.
Pillar 3: Growing Innovative Capabilities
Growing our technical capabilities requires coordination of activities that strengthen LLNL's missions, which means leveraging the collaboration hub in several ways. The AI3 will facilitate the transition of collaboration hub projects to Lab programs, channeling the latter's needs and directions back into the hub and supporting cross-program activities with AI and HPC expertise. Ultimately, to carry out LLNL's institutional AI strategy, we are investing in people, tools, and infrastructure. Learn more about our mission impact, and browse a list of selected high-impact publications.
Working with external partners from industry and academia is crucial for the AI3, for many reasons, and can be a “win” for everyone involved. Most work advancing AI is outside the Lab, and we need to be fully engaged in it. For our part, the Lab can offer important and unique multidisciplinary, funded science applications. We can coordinate multi-partner expertise focused on applications, and we bring visibility and influence with sponsors and government partners. Strong external engagement will help bring new capabilities and tools to Lab programs and missions, as well as build support for LLNL with industry and academic leaders.
We are proud to partner with commercial entities, the ATOM consortium, and academic institutions. We look forward to forming additional collaborations in the coming months. Initial commercial collaborators include the following: