Biden plans to step up government oversight of AI with new 'pressure tests'

President Biden announced sweeping executive actions on Monday to try to establish oversight of the rapidly evolving artificial intelligence sector, setting new standards for safety tests for AI products – as well as a system for federal tests of major systems.

Months in the making, the executive order reflects White House concerns that the technology, left unchecked, could pose significant risks to national security, the economy, public health and privacy. The announcement comes just days ahead of a major global summit on AI taking place in London, which Vice President Harris will attend.

"We're going to see more technological change in the next 10, maybe the next five years, than we've seen in the last 50 years," Biden said during remarks at the White House.

"The most consequential technology of our time, artificial intelligence, is accelerating that change," Biden said.

The order requires companies to share test results for major AI systems

Tech companies currently do their own "red-teaming" of products – subjecting them to tests to find potential problems, like disinformation or racism. The White House has already worked with the major developers on a series of voluntary commitments to red-team their systems by third parties before releasing them.

But Biden's executive order requires the government to set new standards, tools and tests for red-teaming – and requires companies to notify the government and share the red-teaming results for the products that could pose major risks before releasing systems. The power to require companies to do so comes from the Defense Production Act, a Korean-War era law that expands presidential authorities, especially when it comes to national security issues.

"We need to govern this technology. There's no other way around it in my view. It must be governed," Biden said.

In an interview with NPR, White House Chief of Staff Jeff Zients said the executive order inserts "pressure tests" into the AI development process.

"At the end of the day, the companies can't grade their own homework here," Zients said. "So we've set the new standards on how we work with the private sector on AI, and those are standards that we're going to make sure the private companies live up to."

While the Defense Production Act allows for the White House to have some authority over companies' new products, there is still a lack of enforcement power overall — which White House officials say will be up to Congress to legislate.

Biden announced that a bipartisan group of senators, lead by Senate Majority Leader Chuck Schumer, would meet with him at the White House on Tuesday so discuss further actions Congress can take on crafting new laws for AI.

The White House sees the opportunities in AI, but also warns of the risks

Biden said he sees the potential for AI in helping develop new drugs and cures for cancer. But the White House has also expressed concern about the drawbacks, such as "seniors being scammed by voice-cloning technology," Zients said.

"AI can use data — your own personal data — to make social media even more addictive for you or your kids. That's not a good thing. AI systems can use your data to discriminate against a person of color who wants to buy a home. That's unacceptable," Zients said.

The new order asks agencies to look into set up standards for how departments can implement AI in positive ways — like developing new drugs, and assisting teachers in classrooms — but also calls on departments to set standards for making sure AI isn't exacerbating discriminatory practices in housing or loan approval.

But some experts, like Ifeoma Ajunwa of Emory University, said these everyday risks with AI don't get enough of a focus in the executive order.

"The actual present danger is not AI becoming too intelligent. It's more that humans are using AI in ways that are counter to our democratic beliefs about equal opportunity and equal protection," Ajunwa told NPR.

"It does not have any specific mandate regarding enforcement actions regarding AI hiring, to ensure that it's not being discriminatory," she added, noting that use of AI in hiring is already a standard practice.

White House officials have said there are some further enforcement mechanisms beyond the Defense Production Act that are within executive authority, but they also emphasize that regulation must come from congressional legislation.

Recruiting an AI-equipped government workforce

The expansive nature of the order means that federal government workforce will have to be prepared for all the AI-related changes coming, too. Many agencies already have personnel who are designated as leaders on artificial intelligence, a senior White House official told reporters. Those personnel were involved in the process of developing the executive order.

But White House officials, including Zients, said they still needs to recruit more specialized talent.

"The computational capacity of AI is doubling every six months or so. So we need to double down on our efforts to make sure that we are staying ahead of the technology and not chasing the technology," Zients said.

The order calls on the streamlining of visa processes so that immigrants who are skilled in AI can come to the U.S. to study and work on AI development.

Copyright 2023 NPR. To see more, visit https://www.npr.org.