To lead in AI, U.S. needs government investment in R&D, workforce — experts

Officials and an industry expert tried to sort out Wednesday exactly what government’s role should be in helping America become a global competitor in artificial intelligence.

“It’s a global race now to see who is going to be the leader in artificial intelligence,” Bill Dally, chief scientist and senior vice president of research at NVIDIA, said during a panel discussion at the company’s GPU Technology Conference.

Dally said the government needs to invest in research “that will lead to even better AI” and work on developing the workforce.

“People say there’s a huge shortage of programmers today, but actually as AI changes the rules and computers basically program themselves by learning, what we need to be a ‘programmer’ changes,” Dally said. “It’s no longer somebody who writes code — it’s somebody who’s more of a data scientist who curates the data, who understands neural networks [and] can configure them to solve a certain task.”

Panelist Lynne Parker, division director for information and intelligent systems at the National Science Foundation, co-lead the task force that developed the recently released national strategic plan for research and development around AI.

[Read more: White House unveils new goals for AI]

When thinking about government’s role in this space, Parker said it is important to note that R&D informs policy and visa-versa.

She noted that regulators can use R&D as an opportunity to get questions answered that help them develop the best rules.

“There is real research needed in a lot of these areas to help us understand how we address issues of security, issues of privacy,” agreed Alan Davidson, director of digital economy at the Commerce Department.

In addition to working to develop policy at home, Davidson says the department has its eye on the international space — an area also mentioned in the White House’s recent report on preparing for the future of artificial intelligence.

The report recommends that “The U.S. Government should develop a government-wide strategy on international engagement related to AI, and develop a list of AI topical areas that need international engagement and monitoring.”

It also suggests the federal government “deepen its engagement with key international stakeholders,” including foreign governments to exchange information, and collaborate on research and development for AI.

“We are trying to build our muscles within federal government to be engaged in these conversations, to be a good a representative of U.S. industry abroad, because these issues are going to come up,” Davidson said. “They’re coming up in Brussels, they come up in in Beijing, and we want to help develop international frameworks that make sense too.”

Nationally, Dally cautioned against restrictive regulations, advocating instead for regulations “that speed our deployment of AI technologies rather than slam the breaks on until everything is completely understood.”

For example, Dally said that if self-driving cars were deployed now, they would save lives. And he said deploying them now is a state-by-state decision without enough guidance from the federal government.

The Transportation Department recently issued guidelines for the development of autonomous vehicles.

[Read more: Administration asserts role in regulating autonomous vehicles]

“I’m looking to the government to set up an environment which allows us to move forward,” Dally said. “Because our competitors around the world will.”

And the competition is indeed fierce. As Parker noted during the panel, China has surpassed the U.S. in the number of authors on deep learning-related publications.

“That’s an indicator that there’s a lot going on in this space,” she said.

Victor Bennett, senior economist for the White House’s Council of Economic Advisers, also discussed the need to foster domestic competition in the space.

“What we think our role in this whole thing is is setting the rules of the road. We want to make sure that the tools are there for the scientists to be able to be productive, and impactful, and we want to make sure that the rules are there to make sure that the impact is broadly felt across America,” Bennett said.

While government funding is important, Bennett said one little-discussed role the government needs to fulfill is in “competition policy.”

“One area in which AI is somewhat special is the importance of data,” he said. “And so if we can make sure that the data needed to train these algorithms, and to make them better, are broadly accessible, then we can make sure there’s a lot more competition in the AI space, and that no one company comes to dominate.”

He added: “And that’ll be better because it’ll lead to more competition, and it’ll lead to more innovation, and it’ll lead to lower prices, so that Americans broadly can take advantage of these technologies.”