In the international arms race to develop the most advanced portfolio of artificial intelligence tools, the United States risks losing its military advantage and exposing itself to a higher volume of cyber and disinformation attacks if it falls behind its competitors.
Beyond these warnings, the National Security Commission on Artificial Intelligence has cautioned that the U.S.’s role as the top authority on AI research could be in jeopardy, with rivals like China gaining momentum in certain key areas.
“We are concerned that America’s role as the world’s leading innovator is threatened,” the commission wrote Monday in its interim report.
Commission Chairman Eric Schmidt, the former CEO of Google and its parent company Alphabet, told reporters in a roundtable discussion Monday that China currently leads other nations in implementing AI for facial recognition, but the United States retains an advantage in terms of invention and research output.
“When we polled researchers, what they said was that China is a fast follower,” Schmidt said. “In other words, they are benefiting from the globalization of knowledge and they use sources from everywhere, but the best and most original papers are still occurring in the West.”
In order to retain a tactical advantage, the panel of private-sector technologists have recommended that the Trump administration should double down on efforts to recruit the best talent in AI research, and reconsider a proposal to cut to research and development funding by 5% in the fiscal 2020 budget, as well as a 10% decrease in basic research funding.
“Limited availability of federal funding contributes to an accelerating brain drain from academia to industry,” the report states. “This trend damages our ability to train the next generation and influences the direction of research toward more commercially-applied problems. The government must redirect this trend soon.”
Despite proposing overall cuts to federal R&D, Trump administration officials have indicated that it would spare AI research from any of those cuts. In an updated FY 2020 budget request released in September, the White House signaled its intention to double agencies’ investment in AI compared to what they spent three years ago.
Schmidt hinted that the commission’s final report would include recommendations for Congress to make budget changes for AI R&D.
“In theory, we have some possibility of influence, where some of this earlier work may not have had that kind of scale,” he said.
Former Deputy Defense Secretary Robert Work, the commission’s co-chair, said the interim report found “untapped opportunities” for the U.S. to build a nationwide infrastructure for AI research and development. Some of those opportunities include the best minds in AI research working in the U.S. to prevent near-peer competitors like Russia and China from benefiting from those breakthroughs.
“The fact of the matter is our system is the one that is producing the greatest breakthroughs, and we don’t want to give that up, we want to fight for that and make it stronger, not weaker,” Work said.
While the commission has highlighted areas where government and its industry partners should step up their efforts, Work said the interim report doesn’t indicate clear winners and losers in the AI race.
“We try not to have a scorecard. That’s not the intent,” Work said. “The intent is to say, this is a competition that is very, very important for America — for its economic future and its national security future. And we were intent on trying to lay out the things that will allow America to win the competition.”
At a conference Tuesday, the commission will kick off a yearlong effort to gather public feedback for a final report its members expect to complete sometime next year, but without a public release. Instead, Work said the commissioners will send the final report directly to Congress with a set of recommendations for lawmakers to consider implementing.
The commission’s recommendations in its interim closely resemble many of the same conclusions the Defense Innovation Board, which is also chaired by Schmidt, reached in its final report on AI ethics.
Work said the common ground reached by these reports shows a cohesive AI strategy that’s been coming into focus for the past year-and-a-half.
“The fact that most of these different things are coming up with generally the same [lines of effort] — in general, the same initial assessments — is a good thing,” he said. “It indicates the United States is coming to kind of a consensus.”