A panel Congress created to study artificial intelligence’s impact in national security recommends Congress and the Biden administration ramp up AI research and development funding over the next few years for the United States to maintain its tactical edge over its adversaries.
The National Security Commission on AI, in the draft version of its final report to lawmakers, recommends doubling federal R&D spending for AI each year, until it reaches $32 billion in fiscal 2026.
The report also recommends the Defense Department and the intelligence community work together to develop enterprise-wide AI proficiency by 2025.
The commission will issue its final report to Congress on March 1.
Former Deputy Defense Secretary Robert Work, the commission’s vice chairman, said the report urges DoD to reach “military AI readiness” by 2025 through training and education for service members and shared access to AI software-trained models.
The commission also recommends that the Secretary of Defense should established AI readiness performance goals by the end of 2021.
Work said the report’s recommendations will improve DoD readiness, but will require top-down focus from the agency’s top brass and innovation from its combatant commands.
“There is plenty of evidence that the department really is starting to understand how important AI is to their future competitiveness,” Work said in a plenary session the commision held Wednesday.
Demand for AI tools remains high within DoD. Work said DoD’s Joint AI Center so has received 130 requests for help from combatant commands.
“The change agents are here now. It comes in the buy-in from the combatant commanders and also the service chiefs, who are actually actively pursuing autonomous systems … and are looking for decision-making tools, analytical tools [and] readiness tools,” Work said.
Intelligence community AI goals
The commission also sets goals for the intelligence community to adopt AI technologies at scale and maintain a strategic advantage over its competitors.
The report recommends ODNI and DoD create a “red team” dedicated to mitigating the threat of adversarial attacks on AI systems, and urged ODNI to establish an AI integration team with DOD focused on sharing AI methods and data.
The report would also establish a 10-year, $1 billion program of record to provide long-term predictable funding for technologies identified in the technology annex of the National Intelligence Strategy
The report recommends the Office of the Director of National Intelligence work with DoD to establish an AI integration team, focused on maximizing knowledge data and AI model sharing.
The report recommends standing up a chief technology officer for the intelligence community that would develop and monitor IC-wide metrics for AI investments, implementation and readiness, and would ensure maximum sharing and reuse of AI models and code across intelligence agencies.
$32B in AI research by 2026
To scale and coordinate federal R&D funding into AI, the commission proposes a National Technology Foundation focused on incubating AI use cases that have demonstrated early promise into successful commercial technology.
“This requires the NTF to be able to accept very high levels of risk in what they’re willing to invest in. It needs to be empowered to make big bets on innovative ideas and people and funded appropriately to support tech development,” Commissioner Andrew Moore, the director of Google Cloud AI, said. The NTF would complement, but not replace the National Science Foundation, and would lead the “necessary long-term investment in America’s ability to lead in AI and other disruptive technologies.”
The commission supports doubling the AI research and development budget for federal agencies year-over-year until it reaches $32 billion annually.
NSCAI Chairman and former Google CEO Eric Schmidt said AI depends on R&D advancements in related scientific fields, and urged the Biden administration to increase AI research spending at the expense of other federal R&D priorities, as the Trump administration had proposed.
“The White House agreed with our interim recommendation of increasing AI, and they took it out of everything else, which is precisely not what we wanted. I think it’s important we state that this has to be incremental, and it’s not coming out of other science budgets,” Schmidt said.
A bipartisan bill led by Sens. Chuck Schumer (D-N.Y.) and Todd Young (R-Ind.) would rebrand NSF as the National Science and Technology Foundation, and would give the new agency $100 billion over five years to lead investments in AI, machine learning and other emerging technology.
Recruiting and training AI workforce
The commission is also moving forward with its plans for the U.S. Digital Service Academy, a university whose students would agree to serve five-year terms in government service after graduating.
The report also recommends creating a National Digital Reserve Corps that would recruit a cadre of highly skilled digital experts to work at least 38 days a year as short-term advisers, developers and instructors within the government. The commission has narrowed the proposal down to Cabinet-level agencies.
Commissioner José-Marie Griffiths, president of Dakota State University, said the U.S. isn’t currently able to produce the “quantity or quality of domestic talent” needed to support AI research, and must ramp its recruitment of international talent.
“If we are to sustain the U.S. lead, we must cultivate more potential talent at home and recruit and retain existing talent from foreign countries,” Griffiths said.
The commission also recommends creating an emerging and disruptive technology visa category and providing green cards to all students who graduate with a doctoral degree in STEM fields.