America is great at artificial intelligence—and it’s going to get even greater.
So the White House trumpeted Thursday, in the current administration’s first substantial engagement with the technology widely predicted to upend every area of life and society.
At a meeting that mingled industry, academia, and government, the Trump White House framed AI as a path to continued economic dominance over other nations. The exercise also seemed to expose an administration that lags other countries—and even its US predecessor—in developing its views and strategy on AI’s implications and government’s role in managing them.
“America has been the global leader in AI, and the Trump administration will ensure our great nation remains the global leader in AI,” the president’s deputy assistant for technology policy, Michael Kratsios said at the meeting’s opening. He described staying ahead of other countries in AI as “imperative” because of the potential benefits to US industries and their workers.
Kratsios announced that the Trump administration wanted to help AI development by (carefully) giving companies access to some unspecified government data, and that a new select committee would help government agencies think about and use the technology. He also argued that previously announced plans on job training and broadband would somehow help any workers displaced by AI.
His audience then split into discussion sessions mostly focused on economic applications of AI technology. Industry representatives in the eclectic group ranged from top AI researchers at Google and Uber, to executives from United Airlines and Land O’Lakes.
‘We’re very strong believers in diversity and inclusion but it was not a topic that came up in today’s discussion.’
Arvind Krishna, IBM
The White House meeting comes in the wake of major strategy announcements on AI from China, France, the UK, and the European Union. All have pledged significant new funding for AI research, and spoken of the importance of engaging with ethical challenges such as lost jobs, and systems that pick up unsavory values. The US has previously made progress along these lines: Back in 2016, the Obama administration convened a series of meetings around the country that discussed positive and negative impacts of AI and issued two reports on the technology; Obama spoke to WIRED at length about his hopes and fears for AI.
Kratsios didn’t mention that earlier White House activity, or the other countries spinning up AI strategies. But he did seem to make veiled reference to China’s extensive national plan aimed at drawing level with the United States in AI by 2020, and leading the world by 2030. “Command-control policies will never be able to keep up,” Kratsios warned. One way the Trump administration would ensure America wins in AI, he said, is by keeping government intervention in the technology’s development to a minimum.
Attendees described Thursday’s meeting as positive, if preliminary.
Charles Isbell, a professor at Georgia Institute of Technology, says he didn’t hear much new in Kratsios’ opening speech, or closing remarks from Jared Kushner that ended the day. But he came away impressed by the broad mix of different industries present. The discussion sessions he attended on healthcare and supporting innovation in AI were well-informed and freewheeling, quickly diverging from the White House’s preferred talking points.
“It would be fair to characterize what the administration was interested in was thinking what regulations need to go away, and what’s the proper role for government in any regulation,” Isbell says. Although since AI development isn’t meaningfully regulated today, and the White House made it clear it preferred not to regulate, no specific policies seemed imminent.
In a session on healthcare, Isbell and others instead chewed over the importance of machine learning systems being transparent and trustworthy to humans, for example when advising a doctor on how to treat a tricky case. Arvind Krishna, director of research at IBM, says that issue came up in a session on finance, too, where White House representatives appeared to be “listening very carefully.”
White House staff were more actively engaged in discussions about how the US workforce might need to adapt to AI to keep the country more competitive. Isbell heard plenty of talk, for example, about how jobs such as nursing might change as machines become smarter.
‘The development of technology and AI is not a zero sum game.’
Madeleine Elish, Data & Society
Other aspects of the employment challenges of the AI boom got less attention. Isbell would have liked to hear more about how the government could help address America’s tight supply of computer science and machine learning experts. (Trump did direct the Department of Education to spend $200 million more on school STEM programs last fall.) “Without the people needed to do the work you won’t be able to move forward,” he said.
Isbell also heard scant mention of the perception in academia and industry that Trump’s stance on immigration undermines America’s tradition of leading the world by attracting the best technical talent. Nor did Krishna of IBM. “We’re very strong believers in diversity and inclusion but it was not a topic that came up in today’s discussion,” he says.
Thursday’s meeting left some people who contributed to the last, larger White House AI project feeling glum. Madeleine Elish, now a researcher at independent research institute Data & Society, worked on a report that emerged from a 2016 symposium co-hosted by the White House and NYU on the social and economic impacts of AI. She worries that the Trump administration’s approach stokes unhelpful perceptions of what technology policy is for.
“The development of technology and AI is not a zero sum game and it doesn’t have to be positioned as about national superiority,” Elish says. Despite Thursday’s broad-ranging discussions, she doubts the current administration has the inclination or staffing to engage with nuanced questions like how AI systems in government and industry might adversely affect vulnerable groups. Elish notes that the staff of the Office of Science and Technology Policy, which led today’s meeting as it did most AI efforts for president Obama, numbered in the dozens under the last administration but is now a fraction of the size.
In theory, the way is clear for the Trump administration’s nascent AI strategy to evolve. Isbell says the White House suggested there would be follow ups to Thursday event. “This is clearly the beginning of a conversation, not the end,” he says.
Government and AI
This article was syndicated from wired.com