AI Commission Final Report Confirms U.S. Lagging
WASHINGTON — To keep up with technological advancements and provide for enhanced national security, the National Security Commission on Artificial Intelligence was created. Its purpose was to make recommendations to the president and Congress that would “advance the development of artificial intelligence, machine learning, and associated technologies to comprehensively address the national security and defense needs of the United States.”
In March of this year, NSCAI issued its final report concluding that the U.S. is not sufficiently prepared to defend or compete against China, and to some extent Russia, in AI.
“We believe the U.S. has the advantage in AI-related hardware and high-end semiconductor capabilities, advantage over algorithms, and in people… universities and companies still attract top researchers. But China invests aggressively and has the advantage of integration,” Yll Bajraktari, executive director of the NSCAI recently told the Wilson Center.
“They have a strategy pushing — in addition to civilian — military integration,” he warned. “The effort and attention they are putting across all these areas means they might take leadership in AI in less than two years.”
This could be problematic depending on how countries choose to use AI.
Bajraktari said the commission’s final report provided not only high-level suggestions for advancing AI, but also deeper recommendations on national security implementation and putting “AI at the center of what we do on a daily basis.”
As the commission prepares to dismantle on Oct. 1, he summarized its findings in four main areas: leadership, human talent, supply chains, and research.
“How we organize and push these [recommendations] from top to bottom really matters,” Bajraktari explained. “We need to… prepare ourselves better to use AI not just for mundane and bureaucratic elements and other administrative and logistics pieces… but for sensing and understanding, and to be ahead by seeing patterns and observing situations.”
He also focused on developing the workforce of an advanced AI industry.
“People matter,” he said. “This is not only about AI, but also about cultivating a domestic pool.”
In addition to encouraging Americans to study and plan for careers in this field, he added, “We need to be creating a pathway for highly talented immigrants to stay in the country for this work.”
But Bajraktari said that it isn’t only a potential lack of human capital that could derail U.S. AI dominance. Semiconductors and other high-end processors and computational materials are needed in the technology supply chain, and those are increasingly hard to come by without the right partners.
“We’re facing shortages everywhere,” said Bajraktari, though creating and maintaining partnerships with critical allies, like with Taiwan for semiconductor manufacturing, our technical alliance with India, and EU/U.S. tech and trade dialogues are among the strategies to mitigate this.
And while investing in research is paramount, Bajraktari suggested that even with a continued national focus on R&D, “working with like-minded allies and partners” is essential despite the fact that “there is [currently] no simple institution through which the U.S. can engage our partners.”
“On all categories, from preparation, to decision-making, to execution — we need all allies to be interoperable for computing and streamlining of the process… and to establish principles consistent with democratic values,” he said.
While it may be a challenge for the U.S. to encourage multilateral partnerships on these sensitive issues, cooperation beyond joint research and information sharing could result in standards of technical compliance, increased funding for research and technical means, and a greater ethical and responsible use of AI across the world.
For example, AI can be used to give warfighters sense and preparedness options faster — but with an agreement that it will ultimately be left up to human commanders to make decisions — or AI can be allowed to deploy nuclear weapons, which will impact crisis stability.
“Some countries like Russia and China are more willing to use these systems without testing or following norms or laws,” Bajraktari said, which raises a lot of questions about privacy, civil liberties, and safety. “We need to provide an alternative model to China and Russia based on democratic values and norms.”
“We need to discuss how AI can support unmanned and autonomous operations, making our posture more effective, but also more sustainable,” said Bajraktari, warning, “DOD needs to be ready by 2025.”