§9431 National Artificial Intelligence Research Institutes

Summary

Requires the National Science Foundation director, if money is available, to set up a program to give funding to plan, start, and support a network of artificial intelligence research institutes. The Secretary of Energy, the Secretary of Commerce, the NSF director, and other federal agency heads can also give money to eligible groups or partnerships to create and run an institute. An institute is an AI research center that either focuses on a specific sector (for example, health, education, manufacturing, agriculture, security, energy, or the environment) and includes work on ethics, safety, and social effects, or that works on broad AI challenges like trustworthiness or basic AI science. Institutes must partner with public and private groups (such as federal agencies, colleges and community colleges, nonprofit research groups, federal labs, state/local/Tribal governments, industry including startups, and civil society organizations). They must help turn research into real products, support cross-field research and education across multiple schools and organizations, and build the AI workforce, especially from underrepresented communities. Funds can pay to create and share curated, secure, privacy-protected data sets; build testbeds to try AI systems; run research and education projects on social, economic, health, scientific, or national security issues; provide computing, networking, and data access; offer technical help like software engineering; and do outreach to broaden participation. Grants last 5 years at first. Institutes can apply for additional 5-year awards after merit review. Agencies must use a competitive peer-review process with diverse reviewers and avoid unnecessary duplication. The NSF director must create an “Artificial Intelligence Leadership Network” to coordinate the institutes. All funded institutes and subawardees must be based in the United States.