11 November 2019

BREAKING: National Security Commission on AI Releases Interim Report

By Yasmin Tadjdeh

A much-anticipated interim report from the National Security Commission on Artificial Intelligence — which was tasked by Congress to research ways to advance the development of AI for national security and defense purposes — was released Nov. 4.

“How the U.S. adopts AI will have profound ramifications for our economic well-being and position in the world,” Eric Schmidt, the chairman of the commission and the former head of Google’s parent company Alphabet, said during a meeting with reporters in Arlington, Virginia, to discuss the findings. “We've got to get this right. It's not optional.”

The commission — which released its initial report to Congress in late July — identified five fundamental lines of effort to preserve U.S. advantages in artificial intelligence. They include: invest in research and development; apply the technology to national security missions; train and recruit AI talent; protect and build upon U.S. technology advantages; and marshal global cooperation on artificial intelligence issues.

Vice Chairman Robert O. Work, a former deputy secretary of defense who served in the Obama administration, said the release of the interim report marks the end of the commission’s first analytical phase.


“We really worked on what we consider to be the most urgent challenges … and what we thought might be the most transformative opportunities presented by AI for national security,” he said.

The five guiding principles outlined in the report will steer the commission as it moves toward its next analytical phase where it will come up with its final recommendations. The panel is not yet in a position to make final recommendations, suggest major organization changes or identify specific investment priorities, he said.

The group's final report is due to be released in March 2021.

The report was put together by 15 commissioners — which include a mix of experts from government, academia and the private sector — who have been at work since March. The panel so far has attended more than 100 classified and unclassified briefings, 17 working group meetings and four plenary sessions. Its staff has attended more than 200 meetings with industry, Schmidt said.

“We really are talking to everybody,” he said. “We've had very, very strong bipartisan support from Congress — in particular the AI Caucus — and also from the White House. We've also enjoyed very strong support from academia, civil society and the private sector.”

The commission has worked with a number of U.S. government departments and agencies including the intelligence community, he added. Additionally, Work noted the commission collaborated with allied partners such as the United Kingdom, Japan, Canada and Australia.

The commission is considering seven basic principles as it works to prepare recommendations, including: global leadership in AI; adopting artificial intelligence for defense and security purposes; building a shared sense of responsibility for the welfare and security of the American people between the private sector and government; the importance of people; the preservation of free inquiry; the compatibility of ethics and strategic necessity; and the importance of maintaining the United States’ values and rules of law in its employment of AI, according to the report.

The idea that everything the United States does with artificial intelligence has to be in line with American values is one of the most important themes the commission has uncovered during its research, Schmidt said.

“There are lots of hard choices between economic and security interests, between maintaining our openness and protecting our innovation economy from strategic competition, between commercial and national objectives, and then balancing both short-term and long-term considerations,” he said. “These are hard choices, but they're important to get right.”

The national security implications of artificial intelligence are far-reaching, the commission found in the report.

“Adopting AI-enabled systems responsibly, rapidly and at scale will allow national security organizations to understand and execute their missions faster,” the study said. “It will change the way we defend America; how we deter adversaries; how intelligence agencies make sense of the world; and how we fight.”

However, the technology also poses a threat, which could emerge from two areas: what an adversary could do with AI, and what consequences an AI system could have if employed without safeguards, the report said.

Potential threats include: the erosion of U.S. military advantage; risks to strategic stability; the diffusion of AI capabilities; disinformation and the undermining of the nation’s democratic system; erosion of individual privacy and civil liberties; accelerated cyber attacks; new techniques that could open up vulnerabilities; and accidents, the report said.

Forecasting how artificial intelligence will affect national security is no easy task, the document noted.

“The commission’s attempts to predict AI’s impact on national security is like Americans in the late 19th century pondering the impact of electricity on war and society,” the report said. “To move forward with purpose, we need a vision of the AI-empowered future that we seek to achieve, and envision AI futures we want to avoid for America and the world.”

That future will include a world where artificial intelligence is used to “prolong and enrich lives” and provide information on demand, the report said.

However, the government must work to avoid a future where “AI contributes to a world of greater centralized control; empowers authoritarianism; is utilized as an instrument to repress dissent and impose conformity; destroys truth and trust within societies and between states; and is employed in reckless, irresponsible and unethical ways," the study added.

Despite the transformative potential of artificial intelligence, the U.S. government is not making enough investments in the technology, the report said.

“Federal R&D funding for AI has not kept pace with the revolutionary potential it holds or with aggressive investments by competitors. Investments that are multiple times greater than current levels are needed,” the study said.

The document noted that the requested fiscal year 2020 federal funding for core AI research outside of the defense sector grew by less than 2 percent from the estimated 2019 levels.

Over the past five years, federal R&D funding for computer science increased by 12.7 percent, barely sustaining a field in which tenure track positions grew by 118 percent over the same period, the report said.

“The U.S. government knows how to infuse resources into audacious technology projects, as it did for the Apollo space program or the Human Genome Project,” the report said.

However, great power adversaries are working to close the technology gap, the study noted.

"While the Chinese government has made ambitious public commitments to technology mega projects, the United States has returned to pre-Sputnik levels of federal R&D funding as a percentage of GDP," the report said. "Indeed, the trend is going in the wrong direction, with a proposed 5 percent cut to R&D funding — and 10 percent in basic research — in the FY 2020 budget.”

No comments: