The Trump administration will release a national strategy for artificial intelligence in the spring of 2019, according to a senior White House official.
The strategy will focus on setting priorities for investment in AI-related research and development, preparing the U.S. workforce for the economic opportunities and challenges posed by AI, and identifying ways to preserve the U.S.’s technological advantage, said Lynn Parker, assistant director of artificial intelligence at the White House Office of Science and Technology Policy (OSTP), at a Dec. 4 conference.
“Hopefully in early spring you’ll be seeing an updated version of that R&D strategic plan, as well as our process we’ll be using at the federal level for tracking the progress of those investments,” she said.
Parker, who co-chaired the task force responsible for drafting the 2016 National Artificial Intelligence Research and Development Strategic Plan, hinted that the current iteration will largely reflect continuity on the part of the Trump administration.
In September, the National Science Foundation posted the 2016 strategy for public comment. According to Parker, the “overwhelming majority” of feedback received has been positive. Nevertheless, recent technological and political developments may require additional focus on a subset of issues: “Two years is a long time in an emerging technology area,” she said.
DEPARTURES FROM THE STATUS QUO
Clues to the administration’s AI priorities can be found in a May report detailing findings from the White House Summit on Artificial Intelligence for American Industry. The strategy may depart from past federal policy in three areas: reducing regulatory barriers to AI development, retraining the U.S. workforce for a more AI-reliant economy, and limiting technology transfers from U.S. companies to rival nations.
Deregulation has been a centerpiece of the Trump administration’s health care and financial services policies, and it’s likely to apply the same approach to artificial intelligence research and development. “Overly burdensome regulations do not stop innovation – they just move it overseas,” the May report stated.
In the fall of 2017, the White House told the Transportation Department to loosen restrictions on driverless vehicles and on commercial drone technology. Future deregulatory efforts may focus on expanding the amount of consumer data available to companies for the purposes of training AI algorithms.
While the report will offer guidance on preparing high-skilled U.S. workers for jobs in the AI industry, it will also focus on ways to retrain Americans whose jobs may be displaced by automation. According to Parker, the strategy will have a sharper focus on adult education and training. “The technology changes so quickly and your skills can get out of date very quickly,” she said. “We want to provide opportunities for adults to re-skill.”
Finally, the Trump administration is likely to codify some of its recent actions aimed at stemming the transfer or theft of AI-related intellectual property from U.S. companies to competing nations, namely China. In November, the Commerce Department publisheda list of technologies, which included AI tools such as neural networks and deep learning, that could be subject to stricter export controls. The previous month, the Treasury Department announced that it planned to scrutinize foreign investment in U.S. technology firms on national security grounds.
However, it’s likely that any restrictions on U.S. exports will come under fire from U.S. companies such as Alphabet Inc. and Apple Inc., as they attempt to market technologies such as voice-activated devices and self-driving vehicles overseas.
‘TRAINING’ AI ON GOVERNMENT DATA
Two areas where the new AI strategy will likely build on previous efforts are the need for agencies to allocate more money to R&D and the need to share more government data with the public. Parker highlighted the Defense Advanced Research Projects Agency’s pledge to invest more than $2 billion in AI-related R&D over the next five years – an initiative called “AI Next” – as an example of what the Trump administration hopes to see from federal agencies.
In addition to raw funding, the government plans to promote AI research by expanding the number of government datasets available to the public. AI developers can use this data to “train” their algorithms to make more accurate predictions. The strategy may borrow from portions of the President’s Management Agenda, which called on agencies to make “government data accessible and useful through commercial ventures, innovation, or for other public uses.”
The challenge is that releasing this data requires a certain degree of effort on the part of agencies to ensure that the data is in a useable format and that priorities like privacy protection are observed. The 2016 AI strategy cited the lack of “vetted and openly available data sets” as a critical factor limiting the “confident advancement in AI.” The forthcoming strategy could encourage data sharing by aligning it with other presidential management priorities, such as “manag[ing] data as a strategic asset” to improve decision making and accountability.
Between now and next spring, the OSTP’s Select Committee on Artificial Intelligence will take the lead in analyzing the hundreds of comments the government received from industry and the public, and updating federal policy. The committee is chaired by the heads of three agencies most closely involved in AI R&D: France Cordova of the National Science Foundation, Steven Walker of DARPA, and Michael Kratsios of OSTP.
Parker said the committee will also oversee “coordination across all federal agencies in R&D, to make sure that our investments are being coordinated, that we’re not duplicating investments, and that we’re spending taxpayer dollars wisely.”
Chris Cornillie is a federal market analyst with Bloomberg Government.
To contact the analyst on this story: Chris Cornillie in Washington at firstname.lastname@example.org