Over a tumultuous four years, the Trump administration has steadily emphasized the importance of artificial intelligence to American competitiveness. Now President Trump must decide whether to veto what would be the government’s biggest-ever funding and strategy boost to AI.
The National Defense Authorization Act would provide $6.4 billion in federal money over five years for research on AI and its applications, and it would push Washington toward developing a national strategy on the technology.
The bill, approved by both houses of Congress, would increase federal AI spending with $4.8 billion for the National Science Foundation, $1.2 billion via the Energy Department, and $400 million for the National Institute of Standards and Technology.
Martijn Rasser, a senior fellow of the Technology and National Security Program at the Center for New American Security, a strategy think tank in Washington, DC, says the funding is significant.
The bill would also help coordinate the government’s AI strategy, Rasser says, by establishing an office dedicated to AI within the White House Office of Science and Technology Policy. Rasser says this could help guide investment and use of AI, to ensure it is deployed ethically, and to align it with priorities concerning the future of the American workforce.
The defense bill would also create a task force to examine the resources needed for AI researchers. This should lay the foundation for a National Cloud Computing platform for AI research. “That will really help researchers in smaller companies and universities that don’t have the massive resources that the big tech companies have,” Rasser says.
The bill may also help the DOD harness AI more effectively, by giving new acquisition powers to the Joint Artificial Intelligence Center, part of the Defense Department, and by having its head report directly to the secretary of defense.
Supersmart algorithms won’t take all the jobs, But they are learning faster than ever, doing everything from medical diagnostics to serving up ads.
Rasser and some other experts argue that the US government needs to rethink its overall strategy and increase investment in AI research, in order to maintain its leadership and effectively counter China’s rising technological capabilities. It is tricky to tally up total current government spending on AI research, but a report coauthored by Rasser in December 2019 estimated non-defense spending at around $1 billion for fiscal year 2020. The same report recommended that the government increase spending on AI to as much as $25 billion per year.
Some AI researchers are keen to see the bill signed into law. “We cannot afford to fall behind in AI,” says Oren Etzioni, CEO of the Allen Institute for AI, which endorsed the AI portion of the legislation. “Our national security, economic vitality, medical innovation, and scientific progress will depend on it critically in the coming years.”
The defense bill also includes provisions that would require the government to come up with a plan to spend an additional $10 billion per year on advanced technologies such as AI, quantum computing, and 5G wireless services by 2025.
The $740 billion defense spending bill was passed by the Senate and the House last week. The president has repeatedly said he will veto the bill, giving a variety of reasons, including a provision to change the name of military installations named after Confederate officers and the lack of language to alter Section 230 of the Communications Decency Act, which protects tech firms like Facebook and Twitter from liability for content they host.
Trump has 10 days to decide whether to veto the bill. On Sunday, the president tweeted that the defense bill would benefit China without explaining why. All eight of his previous vetoes have been upheld.
Tony Samp, a lawyer at DLA Piper and a former aide to the Senate Artificial Intelligence Caucus, says the bipartisan support for the bill reflects “a recognition that AI is a game-changing technology.” He also points to parts of the National Artificial Intelligence Initiative Act that offer guidance on future AI research, emphasizing the need to account for algorithmic bias and the importance of “trustworthy” AI systems.