WASHINGTON—Technology companies are racing to get ahead of regulators to shape the future of artificial intelligence as it moves deeper into our daily lives.
Companies are already working artificial intelligence, or AI, into their business models, but the technology remains controversial. So
and associations representing Apple Inc.,
Google unit are seeking to set ethical standards, or a sort of code of conduct, through alliances with futurists, civil-rights activists and social scientists.
Critics, however, see it as an effort to blunt outside regulation by cities, states or the federal government, and they question if tech companies are best suited to shape the rules of the road. For the corporations, the algorithms will be proprietary tools to assess your loan-worthiness, your job application, and your risk of stroke. Many balk at the costs of developing systems that not only learn to make decisions, but that also explain those decisions to outsiders.
When New York City proposed a law in August requiring that companies publish source code for algorithms used by city agencies, tech firms pushed back, saying they needed to protect proprietary algorithms. The city passed a scaled-back version in December without the source-code requirement.
“They’re hiding behind trade secrets so you can’t even get a look at what people are doing,” said Ryan Calo, a University of Washington law professor who has advised Congress members on AI issues.
AI, broadly speaking, refers to computers mimicking intelligent behavior, crunching big data to make judgments on anything from avoiding car accidents to where the next crime might happen.
Yet computer algorithms aren’t always clear on their logic. If a computer consistently denies a loan to members of a certain sex or race, is that discrimination? Will regulators have the right to examine the algorithm that made the decision? What if the algorithm doesn’t know how to explain itself?
The Obama administration sought to address these issues. Under Mr. Obama, the Office of Science and Technology Policy issued white papers on the ethical implications on AI. Under Mr. Trump, the office still doesn’t have a director, and its staff is down to about 45, from about 130. A spokesman for the office declined to comment.
For now, the Trump administration has signaled it wants business take the lead. The administration is worried overarching regulation could constrain innovation and make the U.S. less competitive, Michael Kratsios, the deputy in charge of tech policy at the Office of Science and Technology Policy, said at a conference in February. He noted China’s push into artificial intelligence, which it is doing without much ethical quibbling.
“We’re not looking for, or will be endorsing, broad, high level guidelines about AI rules,” Mr. Kratsios said. He called it “key” to create “the regulatory environment that ensures that the next great technologies happen here in the U.S.”
In the past six months, Intel, IBM,
and the Washington, D.C.-based Information Technology Industry Council—whose members include Facebook, Apple and Google—all issued principles on the ethical use of artificial intelligence. In January,
put out an entire book on “Artificial Intelligence and its Role in Society.”
In 2016, some of the biggest tech companies founded an ethics-setting organization called the Partnership on Artificial Intelligence to Benefit People and Society, based in San Francisco. There are many others with at least some industry funding, including Open AI, the AI Now Institute, doteveryone, and the Center for Democracy & Technology.
Proposed rules have ranged from specific guidelines for government use, to requirements that any algorithm be able to explain its process to consumers. Many would apply existing regulations to AI case-by-case, adapting aviation rules to drones or applying privacy protections to personal data in algorithms.
Setting rules is complicated because companies often can’t explain how their more complex systems, called “deep neural networks,” arrive at answers. Last year, University of Washington researchers reported that an algorithm that had learned—by itself—to distinguish between wolves and husky dogs appeared to be doing so by noticing snow on the ground in the wolf pictures, not because of any insight into the animal.
“Google, Facebook, those guys are very interested in performance and making these programs as smart as they can,” said Dave Gunning, who leads a project into making algorithms explain their processes for the government’s Defense Advanced Research Projects Agency.
“Explainability is not as important to them,” Mr. Gunning said. Google and Facebook didn’t respond to requests for comment.
At least one business executive,
founder and chief executive of Tesla Inc. and SpaceX, has cut back his role in the ethical research group he co-founded, Open AI. Mr. Musk quit the group’s board in February. The group said he wanted to “eliminate a potential future conflict of interest” with his role at Tesla, which is developing AI systems. He remains an Open AI adviser and funder.
Some in Congress are taking up the ethical debate. Washington Sen. Maria Cantwell and Maryland Rep. John Delaney, both Democrats, led a bipartisan bill introduced in December to establish a federal advisory council on the technology’s potential impact. There is also an AI caucus on the Hill, started in may, and committees have been holding hearings on AI, algorithms and autonomous vehicles. The European Union has already set regulations for AI algorithms, that are set to take effect in May.
Human-resources software maker Workday, which uses Ai algorithms, assumes that either now, or sometime soon, the U.S. will start to set guidelines.
“Our engagement now is in anticipation that one day there will be government regulation related to AI,” said senior vice president
He said Workday wants to “ensure a regulatory framework that serves society’s needs while allowing the potential of AI, and its benefits, to flourish.”