U.S. lawmakers authorized a slew of national programs that promote artificial intelligence research, development, and deployment, and support efforts to make sure the results are ethical and trustworthy.
What’s new: The 4,500 pages of the National Defense Authorization Act (NDAA), which primarily serves to authorize programs for the U.S. military, includes provisions that promote AI in both civilian and military agencies, and academic institutions, too.
What it says: The NDAA only authorizes these programs; funding will come with further legislation. Among its provisions:
- The National AI Initiative will coordinate research and development across civilian, intelligence, and defense agencies.
- The National Science Foundation will begin planning a National Research Cloud, an aggregation of processing and data resources to be made available to academic and nonprofit researchers. (Fei-Fei Li described the National Research Cloud in our special New Year issue of The Batch.) The NSF will also build AI research institutes focused on health care, manufacturing, and other sectors; study the impact of AI on the nation’s workforce; and sponsor competitions that promote innovation.
- The National Institute of Standards and Technology will create a framework to grade AI systems on trustworthiness and define related terms like explainability, privacy, and transparency. The agency also will formulate privacy and security standards for training datasets, data management, and AI hardware. The Defense Department must ensure that any AI it acquires was developed “ethically and responsibly.”
- The Joint AI Center, a military organization launched in 2018, will report directly to the Deputy Secretary of Defense, giving the Pentagon leadership more direct control over its research and development priorities. The center’s biannual report must describe its work developing AI standards and its collaborations with other agencies.
Yes, but: Some of these programs, such as the NSF’s AI research institutes, will cost money that Congress has yet to appropriate. Russell Wald, director of policy at Stanford’s Institute for Human-Centered Artificial Intelligence, told The Batch he’s optimistic that funding will be allocated where it’s needed.
Behind the news: President Donald Trump vetoed the NDAA in December, saying he wanted it to include a repeal of Section 230 of the Communications Decency act, which protects internet companies from legal liability for user-generated content on their sites. Congress overrode the veto and passed the bill into law on New Year’s Day.
Why it matters: The U.S. is a global leader in AI innovation and home to more big AI companies than anywhere else, but its government has lagged in issuing a comprehensive AI strategy. Directives like the National Research Cloud would give a healthy boost to AI researchers in many areas, and the impact likely would ripple across the world.
We’re thinking: This bill is a major step forward in U.S. support for AI. We’ll keep our fingers crossed that the necessary funding comes through.