A White House-led task force is asking Congress to fund its plans to make the federal government’s artificial intelligence resources more accessible to a broader community of researchers.
The task force behind the National AI Resource (NAIRR), a federal AI data-and-research hub, is telling Congress in a final report released Tuesday that it could reach initial operating capability within 21 months, if the project receives enough funding.
The final report estimates the NAIRR will need about $2.6 billion in congressional appropriations over its first six years. Most of the funding would go to multiple agencies contributing AI resources to the NAIRR.
The task force based its funding request on the cost of advanced computing and other resources expected to be in higher demand, as the AI research and development community grows.
The NAIRR, as envisioned by the task force, would connect the research community to the computational data and testbed resources essential to AI R&D through an online portal.
A NAIRR task force member told reporters that AI is driving scientific discoveries and economic growth across a range of sectors, “but at the same time, it’s raising new challenges related to its ethical and responsible development.”
“While AI research and development is rapidly advancing in the United States, access to the computational and data resources that drive the cutting edge of AI remains primarily limited to those who are working at large tech companies and well-resourced universities,” the task force member said Tuesday. “This a really unique situation where the ability to pursue technological innovation in a key technology is so exclusive to private companies.”
By lowering the barriers to participate in AI research, and allowing more researchers to get involved in AI development, the NAIRR task force expects the community will resolve some of the emerging challenges around ethical AI use.
AI research relies on access to large amounts of computational power and data, which often limits this research to the federal government, large technology companies and top universities.
“We see that realizing the potential of AI to solve big, societal problems depends on bridging this access divide, and cultivating an ecosystem that looks like America and works for every American,” the task force member said.
The final report outlines four main goals for the NAIRR — to spur innovation, increase the diversity of AI talent, increase capacity for AI research in the U.S. and advance trustworthy AI.
The task force recommends an external, non-government entity handle the NAIRR’s day-to-day operations, but the National Science Foundation would serve as its “administrative home.” NSF would also provide funding and oversight to the NAIRR.
A steering committee with other federal agencies as members would set the NAIRR’s strategy and goals, and would decide which additional AI resources are added to the NAIRR.
The final report doesn’t name the non-government entity that would run the NAIRR. Instead, OSTP and NSF propose working with the steering committee to stand up a program management office, which would select the entity after submitting a request for proposal.
The non-government operating entity behind the NAIRR would receive about $55-$65 million every year to support the coordination and management of NAIRR activities, according to the task force’s funding request. Another $5 million a year would go toward oversight and evaluation of the NAIRR’s performance.
Another task force member said the NAIRR will be comprised of new and existing resources, some of them owned by private-sector third parties.
The task force recommends the non-government entity behind the NAIRR develop standards and criteria for the types of additional AI resources it will accept. It also directs that entity to adhere to an AI framework that NIST expects to release on Friday.
“We see the rollout of these two reports on the week as really showcasing the dual priority of advancing AI innovation, and doing so in a manner that mitigates risk and advances best practices in responsible and trustworthy AI,” a task force member said.
The task force also recommends the NAIR’s non-government operator, as part of the Biden administration’s Blueprint for an AI Bill of Rights, should require researchers to complete ethics training before gaining access to its resources.
The NAIRR task force also points to the recently passed CHIPS and Science Act of 2022 another reason to prioritize equitable access to national AI research.
The 2020 National AI Initiative Act mandated the formation of the NAIRR task force, with the White House Office of Science and Technology Policy and NSF as its co-chairs.
OSTP and NSF launched the NAIRR task force in June 2021. Since its launch, the task force held 11 public meetings and issued two requests for information on how the NAIRR should operate.