Dive Overview:
- The U.S. Department of Energy said Tuesday:Roadmap“Implementing the frontiers of artificial intelligence for science, security, and technology; or Fast, The initiative aims to develop U.S. AI capabilities for scientific discovery, energy research, and national security.
- The program would be authorized by a five-year budget of $2.4 billion per year included in the bipartisan Department of Energy AI Act. Fast According to the agency, the plan aims to leverage the capabilities of DOE and its 17 national laboratories to “build the world's most powerful integrated scientific AI system.”
- “We need sound, accountable governance and compliance structures guiding the use and application of AI,” Paul DeCotis, a senior partner at business and technology consulting firm West Monroe, said in an email. “Government leadership, with public input, is welcome.”
Dive Insights:
FASST will “transform the vast repositories of scientific data generated at DOE's user facilities into AI-enabled computers and build the next generation of energy-efficient AI supercomputers,” the department announced Tuesday.
In April, the DOE released two reports concluding that AI could help manage the U.S. power grid, including by reducing emissions and lowering costs, but also warned that “naive” deployment of AI could expose the country to a variety of risks, including cyberattacks, physical attacks on the power grid and supply chain compromises.
AI is “a transformative technology that will help drive breakthroughs in energy technology and strengthen our national security,” Energy Secretary Jennifer Granholm said in a statement on July 16. “FASST further strengthens DOE's role as the national steward of advanced supercomputing and research infrastructure.”
According to a fact sheet distributed by DOE, the effort will focus on four related areas:
- Provide AI-ready datasets to partners in government, industry, and the scientific community to train, test, and validate AI models.
- Build next-generation, energy-efficient, AI-enabled supercomputing platforms and algorithms that leverage machine learning and digital infrastructure-driven scientific computing.
- Accelerating discovery in all areas of science and;
- It will revolutionize how DOE accomplishes its science, energy, and security missions.
“Scientific discoveries accelerated by AI could lead to cheaper batteries for electric cars, breakthroughs in nuclear fusion energy, new cancer drugs, and help ensure national security,” the agency said.
Senators Joe Manchin (Ind./Va.) and Lisa Murkowski (R-Ala.) introduced legislation to advance the FASST initiative on July 10. The bill requires the Secretary of Energy to report annually to Congress on the initiative's program's “progress, findings, and expenditures” and sets a budget of $2.4 billion per year over five years.
“As AI technologies take the world by storm, the United States must respond quickly and effectively before our adversaries do,” Manchin said in a statement. “Leveraging existing research facilities and scientific expertise in AI rather than starting from scratch will protect taxpayer dollars and allow us to act quickly.”
The bill also establishes a network of AI research clusters building on DOE's existing infrastructure, calls for an AI “risk assessment and mitigation program” to assess security risks, and directs the Federal Energy Regulatory Commission to begin rulemaking on the use of advanced computing to speed up the interconnection queuing process. The bill also directs DOE to study the growing energy demands of data centers and AI.
The Electric Power Research Institute released a report in May stating that data centers could consume 9% of U.S. electricity production by 2030, roughly double current levels, and that AI queries require roughly 10 times more power than traditional internet searches.
“AI can be used to reduce energy use in data centers and advanced manufacturing, identify new ways to process data and information and streamline operations,” West Monroe's Decotis said. “The Defense Department's work should establish a standard for how AI capabilities and uses can spur innovation and discovery in science, energy and natural security.”
