Visionary report unveils ambitious roadmap to harness the power of AI in scientific discovery

Innovations in artificial intelligence (AI) are rapidly shaping our world, from virtual assistants and chatbots to self-driving cars and automated manufacturing. Seizing on the potential of AI to transform science, the nation’s leading experts in science and technology have released a blueprint for the United States to accelerate progress by expanding its capabilities in AI and big data analysis.

A new report, the product of a series of workshops held in 2022 under the guidance of the U.S. Department of Energy’s (DOE) Office of Science and the National Nuclear Security Administration (NNSA), lays out a comprehensive vision for the Office of Science and NNSA to expand their work in scientific use of AI by building on existing strengths in world-leading high performance computing systems and data infrastructure.

The sessions brought together more than 1,000 scientists, engineers and staff from DOE labs, academia and technology companies to talk about the rapidly emerging opportunities and challenges of scientific AI.

The report identifies six AI capabilities and describes their potential to transform DOE/NNSA program areas. These range from control of complex systems like power grids to foundation models like the large-language models behind generative AI programs such as ChatGPT. The report also lays out the crosscutting technology needed to enable these AI-powered transformations.

“Advances in AI could dramatically change the way we pose and solve scientific problems,” said Rick Stevens, associate laboratory director of Computing, Environment and Life Science at DOE’s Argonne National Laboratory (ANL), who co-led the workshops and report development. “There’s so much untapped potential for AI innovations to both drive new scientific discoveries and speed up the pace of those discoveries. DOE is the natural leader for developing trustworthy, responsible AI to support our nation’s energy transition and nuclear deterrence efforts.”

The report describes scientific “grand challenges” where AI plays a major role in making progress toward solutions. These include improved climate models, the search for new quantum materials, new nuclear reactor designs for clean energy and more.

Lawrence Livermore National Laboratory’s (LLNL) Advanced Simulation and Computing (ASC) Program Director Rob Neely, who co-led the workshop series, said the event and subsequent report came at a “optimal time for DOE and NNSA to begin to sharpen our strategy, aimed at taking advantage of the incredible pace of artificial intelligence.”

“The power of AI is now on the minds of everyone with the emergence of Large Language Models [LLMs] like ChatGPT and the immense power they hold, and it will be imperative for the U.S. to take a leadership role in the research and application of this technology,” Neely said. “DOE is truly uniquely positioned to make that a reality with our abundance of high-powered compute, our ability to put together large teams and our mission of science and national security. LLNL’s ASC program is confident that we will build upon years of research to avail this once-in-a-generation shift in how science and knowledge will be done.”

LLNL is making a significant investment in AI, particularly through its application of “cognitive simulation” — combining AI-driven workloads with high-performance computing — to NNSA’s stockpile stewardship programs, where they are improving models for stockpile modernization, inertial confinement fusion experiments and other key scientific efforts.

Neely added that LLNL’s deployment in 2024 of the exascale-class El Capitan, which will be amenable to AI-assisted data analysis, will be a “game-changer” and a “powerhouse system for traditional modeling and simulation, as well as one of the only systems on earth capable of training large AI models in a classified environment.”

DOE is uniquely positioned to extend the United States’ global leadership in science, energy and security, the report notes. DOE operates the world’s most powerful supercomputers — including the new exascale systems Frontier, Aurora and El Capitan — encompasses the world’s largest collection of experimental facilities and employs more than 50,000 people at its 17 national laboratories.

In addition to ANL and LLNL, four other DOE national laboratories led the AI workshop series and resulting report: Lawrence Berkeley National Laboratory, Los Alamos National Laboratory, Oak Ridge National Laboratory and Sandia National Laboratories.