Senators seek to advance AI capabilities at national labs

Two senators are working to usher more artificial intelligence systems and technologies into the U.S. national laboratory network through new legislation introduced last week.

Sens. Joe Manchin, I-WV, and Lisa Murkowski, R-AK, introduced the Department of Energy AI Act to expand AI computing capabilities across the 17 national laboratories, with a focus on deploying frontier — or leading-edge — AI models to help researchers visualize the myriad, and often classified, datasets in multiple scientific domains.

“As AI technology takes the world by storm, the United States needs to meet the moment quickly and effectively before our adversaries do,” Manchin said in a press release. “The DOE and its network of National Laboratories are ready and able to bring our nation to the next level of scientific discovery and global competitiveness through the innovation of safe and responsible AI.”

One key component of the bill authorizes the Frontiers in Artificial Intelligence for Science, Security, and Technology, an internal initiative within Energy that unlocks funding to bring AI capabilities into multiple department domains. Authorizing this program would let Energy further develop its workforce, data and computing capabilities, and expand its Exascale Computing Project partnerships with industry entities. The bill authorizes $12 billion in spending over five years to launch the FASST program.

The bill’s other stipulations include the establishment of AI risk evaluation and mitigation programs to ensure AI and machine learning technologies are deployed safely. It also asks the secretary of Energy to create a strategic plan with short- and long-term goals required for the advancement of AI applications within science, energy and national security missions. 

Some national laboratories would also be selected on a competitive basis to host a minimum of eight multidisciplinary AI Research and Development Centers which would feature participation from national lab staff, academia and industry. The primary focus of these centers would be developing and deploying AI softwares and components to ensure that frontier AI research methods are “well-suited” for key federal research missions. 

Per the bill, each of these centers would be authorized to receive no less than $30 million in annual funding between five to seven years of operation, with renewal opportunities.  

The bill also pays attention to the potential for AI systems to work with sensitive research areas, particularly related to the U.S. stockpile and other national security realms. It calls for the Energy secretary to create a taxonomy of risks related to using AI in certain classification levels, as well as the generation of nuclear weapons; the creation of cyber and malware threats; the generation of radiological, nuclear, biological and chemical threats; and other hazards to public safety. 

If enacted, the bill also requires the Federal Energy Regulatory Commission to begin a rulemaking procedure that revises current language in Large Generator Interconnection Procedures with the goal of incorporating AI and advanced computational capabilities into electrical grid infrastructure management. 

Within the Department of Energy, leadership would be required to submit a report to Congress on the outcomes of AI in national labs, along with a separate document to assess the growth of large scale data centers and the impact of their computing needs on energy consumption. 

Cultivating a strong workforce to meet the needs of an updated AI infrastructure is further emphasized within the bill text, beyond the initial scope of the provisions in FASST. Manchin and Murkowski ask for no less than 10%of all allocated federal funds for each established Center to go toward education and training opportunities in STEM fields.