New Session Sparks New Priorities for Senate AI Caucus
After steering several AI policies in 2020, the nascent group is embarking on a new agenda.
As the Senate Artificial Intelligence Caucus edges up on its two-year anniversary, members’ sights are set on carrying out an array of new technology-centered policies they collectively helped push forward in the last Congressional session.
“We have our work cut out for us as we ensure all eight bills that became law in 2020 are implemented swiftly and effectively,” Sen. Rob Portman, R-Ohio, told Nextgov last week.
Portman co-founded Congress’ younger bipartisan AI caucus in early 2019. In recent, separate discussions, he and a Senate aide briefed Nextgov on some of those freshly passed legislative measures, and the novel group’s plans for early 2021.
A Busy Beginning
Now an invisible constant underpinning services commonly used by most in the U.S., AI makes up a wide-ranging field that’s constantly evolving and hard to define. Adoption varies across federal entities, but the technology is used by many agencies and has been making major waves across commercial industries.
A range of government-led AI initiatives emerged to drive forward understanding and strategic use of it over the last decade, including a National Artificial Intelligence Research and Development Strategic Plan and companion report from former President Barack Obama’s administration in 2016. And in mid-2017, lawmakers in the House of Representatives launched Congress’ first AI caucus, noting that its ultimate goal would be “to inform policymakers of the technological, economic and social impacts of advances in AI and to ensure that rapid innovation in AI and related fields benefits Americans as fully as possible.”
Multiple AI-related pursuits were also prioritized by former President Donald Trump, who in February 2019 signed an executive order establishing what it called the American AI initiative. Much like Trump’s national quantum initiative, the effort was meant to advance the U.S. tech workforce and research landscape. A month later, in March, Portman—alongside Sen. Martin Heinrich, D-N.M.—unveiled the second AI caucus.
“The House has had an AI caucus for a few years now. The Senate has not played in this space, until now—just a Congress ago,” a Senate aide said. “But I feel that we’ve really stepped up and now we're kind of actually driving the conversation in Congress on AI.”
When the Senate’s AI-interested group was established, members expressed plans to learn about the blossoming technology and produce “smart policy that balances AI’s risks and rewards to ensure the competitiveness of the U.S. economy, while maintaining important ethical standards.” The aide explained that since then, Portman helped craft and put forth “11 different pieces of legislation on AI,” much of which have become law, particularly in late 2020.
Tucked into the more than 5,000-page National Defense Authorization Act for 2021—that Congress overrode a veto from Trump for the first time to pass strikingly close to deadline—were versions of four pieces of that AI-aligned legislation.
“I’m particularly glad that the National AI Research Resource Task Force Act is now law because this measure lays the groundwork to build a national cloud computer so that we can democratize access to supercomputing power for researchers across America, especially those in the middle of the country,” Portman said. “This will make America even more competitive on AI’s world stage.”
That bill mandates a new group of technical experts from public, private and academic sectors to connect and prepare a comprehensive roadmap for how America can design, launch, govern, and sustain a national research cloud. More than a dozen entities, including Stanford University and Google, expressed support for the call.
And its roots also partly trace back to a report from the National Security Commission on AI, which articulated such a need.
“We want to make sure that we're able to maintain parity at least and continue to not totally get lapped by China on supercomputer power. But the other thing is, we want to democratize access to supercomputing power,” the aide said. “You shouldn't have to live in Silicon Valley or work in the national labs to have access to supercomputer power. So, we feel like this is going to be a real game-changer, because we're making this big federal investment that offers a lot of opportunity for researchers in the industrial heartland, in particular, and across the country, of course.”
The bill also represents a push to “move the ball in new areas,” just as another piece of newly passed legislation in the NDAA—the AI Initiative Act—enables new research and development, funds billions in federal investments over the next half-decade and calls for the production of a broader AI national strategy. The aide noted that the caucus “really worked hand-in-hand” with colleagues in the House Science Committee to move the legislation forward over many months. The text essentially codifies the national AI initiative.
Provisions are also incorporated requiring the creation of the National Artificial Intelligence Initiative Office, which the Trump administration announced in its final days.
“This office will be at the center of developing our national AI strategy and I look forward to working with them to develop that strategy,” Portman said.
The Republican lawmaker added that he’s “also proud to see [his] Deepfake Reports Act become law as part of the defense funding bill,” noting that, “[i]n recent years, we have seen an incredible proliferation of false information online.”
Deepfakes are hyper-realistic but AI-manipulated forms of media that make it look like the content’s subjects did or said things they did not. Such videos, audio and images have grown increasingly viral in recent years, including some targeting political leaders. And as America grapples with vastly shared, complex contemporary conspiracy theories and sophisticated disinformation campaigns, certain Congress members want to hone in more aggressively on this potential means for disseminating falsities to the masses. One of those individuals is Portman, who called deepfakes a “growing element of that misinformation ecosystem.”
“[A]nd that’s why we need to start doing serious research into the threats posed by deepfakes and how policymakers should respond,” he added.
Specifically, the manipulated media-focused legislative measure requires the Homeland Security Department to study deepfake-generating technologies, how the content is used by both foreign and domestic entities, and potential countermeasures to ultimately inform and improve Congress’ grasp of the nascent, potential tech-boosted threats.
“What's unique about the report is that there had been some previous reporting requirements in the 2020 NDAA that were just focused on the military—and we wanted to broaden the aperture,” the Senate aide said. “We want to capture not just what our adversaries are doing, we want to capture what's going on in the information ecosystem in the United States, particularly with respect to threats to cause harm, threats to people's civil rights and those types of things.”
Portman and his caucus colleagues also helped steer the passage of several other AI-driving policies last year, the aide noted, adding “I mean, in 2020, we put hundreds of hours on these AI bills to get them done.”
To the official, the fact that the bulk of the latest measures to be passed came in the NDAA and recent COVID-19 omnibus package, is partially a “symptom of gridlock.”
“For these big bills, the NDAA is kind of the place, just like the omnibus is the place—and a lot of these bills do have a national security component,” the aide said. “So, I think people felt that it was appropriate to include it in there.”
AI isn’t new, but the realm continues to gain modern Congressional enthusiasm.
“There's universal recognition at this point of some of the challenges that China poses to the United States, to American competitiveness, in a number of sectors—and obviously AI is one of those,” the Senate aide noted. “And so I think we're kind of in a special moment where everyone has pretty much figured that out now, and wants to do something about it.”
While the new year and Congressional session have only just begun, Senate AI Caucus members are in the market for new technology-pushing ideas. And as Portman previously mentioned and the aide additionally emphasized, these early months will involve oversight and implementation on all that just became law to ensure all the weighty, AI-related aims are met.
“I also believe that Congress should begin to tackle issues related to the future of work,” Portman said. “The pandemic has revealed big changes in our economy and technology has a big role in shaping the nature of work going forward.”
Another initial plan for 2021 is to expand the number of lawmakers involved in the caucus. Presently eight members representing both sides of the political aisle are on board.
“We will do another membership drive early in the Congress here,” the aide said. “I think in terms of attitudes, AI is really the last great non-partisan, or even trans-partisan issue. ... We would not have been able to accomplish all these bills if we didn't have this great bipartisan consensus around how important this stuff is—and so, I don't see that going away.”
Portman also echoed that sentiment.
“We have a great bipartisan group of members, and I look forward to increasing our membership this year,” he said. “Unlike other policy issues, there is no Republican or Democratic position on AI, and as a result we have been able to work very collaboratively across the aisle to secure these victories for our country.”