FYI: Science Policy News

Senate AI Blueprint Proposes ‘Emergency’ R&D Surge

MAY 15, 2024
A new bipartisan blueprint endorsed by the Senate majority leader proposes using “emergency” appropriations to ramp up non-defense AI R&D spending to at least $32 billion per year, with some of the money going to broader priorities such as implementing the CHIPS and Science Act.
Mitch Ambrose headshot
Director of FYI
Bipartisan Senate AI Working Group Schumer Press Conference May 5 2024

Senate Majority Leader Chuck Schumer (D-NY) speaks at a May 15 press conference on artificial intelligence policy.

(Bill Clark / CQ Roll Call)

Today, a bipartisan quartet of senators released a blueprint for artificial intelligence policy that proposes Congress rapidly ramp up federal spending on AI R&D using “emergency” appropriations.

The group is led by Senate Majority Leader Chuck Schumer (D-NY), a key architect of a similar emergency infusion for semiconductor manufacturing and R&D provided by the CHIPS and Science Act of 2022. The group’s other members are Sens. Todd Young (R-IN), Martin Heinrich (D-NM), and Mike Rounds (R-SD).

The senators propose that Congress surge funding for non-defense AI R&D programs across the government to at least $32 billion per year, matching the level proposed in 2021 by the National Security Commission on AI (NSCAI). The commission estimated that federal agencies spent about $1 billion on such R&D in fiscal year 2020 and proposed that Congress double that figure each year over five years.

So far the government has not come close to matching that pace. Non-defense AI R&D spending reached about $3 billion in fiscal year 2023 with about $1.8 billion of the total considered to have a “core” focus on AI, according to the interagency group that compiled the figures. With spending caps in place for the subsequent two fiscal years, further increases will be very difficult to achieve absent special spending measures.

Accordingly, the group calls for Congress to use emergency appropriations to “fill the gap between current spending levels and the NSCAI-recommended level.” It also identifies priority programs for the funds to be spent on, such as:

  • A cross-government AI R&D effort that spans the Department of Energy, National Science Foundation, National Institute of Standards and Technology, National Institutes of Health, NASA, and other relevant agencies;
  • An “AI-ready data” initiative that has a focus on “fundamental and applied science, such as biotechnology, advanced computing, robotics, and materials science";
  • Efforts authorized by the CHIPS and Science Act that have not been fully funded, including but not limited to NSF’s education programs and its Directorate for Technology, Innovation, and Partnerships, DOE’s advanced computing and microelectronics programs, and the Commerce Department’s regional technology development hubs;
  • NSF’s National AI Research Resource and its National AI Research Institutes;
  • “AI Grand Challenge” programs that focus in part on developing applications to “fundamentally transform the process of science, engineering, or medicine";
  • NIST’s AI programs as well as construction projects to “address years of backlog in maintaining NIST’s physical infrastructure"; and
  • A joint NIST-DOE test bed to “identify, test, and synthesize new materials to support advanced manufacturing through the use of AI, autonomous laboratories, and AI integration with other emerging technologies, such as quantum computing and robotics.”

At a press conference today, Schumer emphasized his intent to pursue emergency appropriations and said a similar scale of spending would be needed for defense programs focused on AI.

“We do believe it’s emergency funding. We’re going to figure out the best way to get it done,” Schumer said.

“We do intend that there will be a large defense expenditure as well, probably about as large as the non-defense,” he later added.

In their blueprint, the senators recommend funding efforts to mitigate threats from AI-enhanced biological, chemical, and nuclear weapons, including through testbeds and model evaluation tools developed by DOE.

They even broach the possibility of creating a new classification category for AI analogous to the Restricted Data category DOE uses for nuclear weapons information. Specifically, the senators recommend Congress “develop a framework for determining when an AI system, if acquired by an adversary, would be powerful enough that it would pose such a grave risk to national security that it should be considered classified, using approaches such as how DOE treats Restricted Data.”

At the same time, they call for increasing collaboration on AI with “like-minded” allied countries, including through legislation that establishes “international AI research institutes.”

The blueprint briefly endorses increasing immigration pathways for people with STEM expertise. The senators encourage committees with jurisdiction over immigration policy to “consider legislation to improve the U.S. immigration system for high-skilled STEM workers in support of national security and to foster advances in AI across the whole of society.”

In the press conference, Schumer said the blueprint is intended to “supplement, not supplant” the work of congressional committees. “We always knew we would have to go to the committees to get the specifics done,” he said.

Any initiative will also need buy-in from the House. Schumer said he intends to raise the subject soon with House Speaker Mike Johnson (R-LA), who established his own bipartisan taskforce on AI in February.

Related Topics
More from FYI
The centers will aim to improve the durability and energy efficiency of microelectronics.
Many federal research facilities are operating beyond their planned lifespan and are in poor condition, according to a new cross-agency assessment.
The bill allows the energy secretary to issue waivers but aims to wean the U.S. off Russian nuclear fuel.
The bill proposes new R&D programs focused on next-generation radar, atmospheric rivers, flooding, and aviation weather.

Related Organizations