Avoiding Past Mistakes: Becoming AI First Movers
While Congress and the Executive branch squabble over impeachment, the budget, and the United States–Mexico–Canada Agreement (USMCA), there are thankfully some within government keeping their eye on America’s long-term interest in Artificial Intelligence (AI). As a neophyte technology, artificial intelligence, cannot be perfected or implemented overnight. If it is to win the Artificial Intelligence arms race with China, the U.S. government (USG) must become a first-mover in this space. Accordingly, what is needed is a forward-leaning, whole-of-government approach to AI. Fortunately, the Trump Administration, DoD, and the AI Caucuses in both the Senate and the House are all positioning the U.S. to win this AI arms race.
Historically, the USG is traditionally reactive, only changing its philosophies after a calamity. For instance, with the 1947 passage of the National Security Act creating the CIA, Air Force, and the National Security Council, the USG acted in response to failures exposed during WWII, including lack of coordination amongst government agencies and the inability to manage foreign policy bureaucracy. In 1986, following fiascos in Operation Eagle Claw and Operation Urgent Fury, Congress implemented the Nunn-Cohen Amendment creating United States Special Operations Command (USSOCOM) and the Goldwater-Nichols Act streamlining the military chain of command. Yet another example of the USG’s reactive tendency is the passage of the 2004 Intelligence Reform and Terrorism Prevention Act (IRTPA), in response to the lack of coordination amongst America’s national intelligence and law enforcement agencies before 9/11 and the failure to find Iraqi weapons of mass destruction. In response, IRTPA created the Office of the Director of National Intelligence and reorganized the Intelligence Community to meet similar and future challenges.
While the USG is moving at a glacial pace compared to China in its AI strategy, there is also something to be said for moving in a measured, sobered manner versus recklessly dumping resources into unproven and opaque ventures. The U.S.’ first push toward solidifying AI strategy began with the Obama Administration 2016 AI Research and Development Strategic Plan. It would, however, take almost another two years until the Trump Administration submitted its own updated Artificial Intelligence strategy. The “American AI Initiative,” Executive Order 13859, released in February 2019, “implements a whole-of-government strategy in collaboration and engagement with the private sector, academia, the public, and like-minded international partners.”
Further, “it directs the Federal government to pursue five pillars for advancing AI: (1) promote sustained AI R&D investment, (2) unleash Federal AI resources, (3) remove barriers to AI innovation, (4) empower the American worker with AI-focused education and training opportunities, and (5) promote an international environment that is supportive of American AI innovation and its responsible use.” This initiative serves as a symbolic document, demonstrating the Trump Administration takes the future of AI quite seriously. In-line with the free market approach underlying the “American AI Initiative,” however, it does not allocate any new resources to furthering the prior strategy, instead, it prioritizes federal agencies to proactively spend on AI.
In conjunction with the American AI initiative, the Trump Administration also released the DoD AI Strategy in February 2019. The document paired well with the 2018 National Defense Strategy, and focuses on both increasing AI funding while also accelerating the speed at which AI investments are made. A critical component of this document calls for DoD to partner with industry and academia to find new opportunities to implement its AI strategy.
As part of this overall Defense re-prioritization around AI, the Joint Artificial Intelligence Center (JAIC) stood up "to provide a common vision, mission and focus on driving department-wide AI capability delivery." Its four lines of effort include: (1) accelerate delivery and adoption of AI capabilities across DoD, (2) establish a common foundation for scaling AI’s impact, (3) synchronize DoD AI activities, related AI and machine-learning projects to ensure alignment with the National Defense Strategy, and (4) attract and cultivate a world-class AI team.
Moreover, in an effort that may make it easier for the USG to work with academia and the private sector, the Defense Innovation Board released its AI Ethical Guidelines in early-November 2019. These guidelines organize around the following concepts: responsibility, equitability, traceability, reliability, and governability. These ethical guidelines aim to help DoD build trust, transparency, and dependability with its AI systems.
The third major actor in this AI push is the legislative branch, headed by both the House and Senate AI Caucuses, and which serves to allocate resources towards AI. These caucuses - launched in 2017 and 2019, respectively - actively push AI legislation in both houses, with the Senate being the more legislatively active of the two branches. In my talks with Congressional staffers on the House and Senate AI Caucuses, I’ve learned that many Congressmen are actively tied into the AI ecosystem. In fact, many congressional staffers on and outside the AI caucuses attend weekly meetings with industry and academia to learn about AI and transmit this information back to their Congressmen. As one AI Caucus staffer framed it, “we are not waiting for a crisis to occur. Rather, we are working to ensure American competitiveness abroad through a federal approach to AI.”
Included in this federal approach is the 2019 Artificial Intelligence Initiative Act, introduced by the co-founders of the Senate Artificial Caucus, Senators Portman and Heinrich. This act was purposely released alongside the American AI Initiative, and authorizes “$2.2 billion federal investment over five years to build an AI-ready workforce.” In May 2019, the Senate AI Caucus, plus Senator Kamala Harris, reintroduced the Artificial Intelligence in Government Act to articulate the rules by which the federal government can internally use AI. The Caucus also recently passed legislation on Deepfakes, directing the Department of Homeland Security (DHS) to both study Deepfakes and develop countermeasures and regulations for this specific AI technology.
In addition to these initiatives, the Senate AI Caucus also approved provisions to the Senate-passed National Defense Authorization Act (NDAA) on the Armed Forces Digital Advantage Act, as well as amendments on explainable AI and China’s AI capabilities. Principally, the Armed Forces Digital Advantage Act, acting on recommendations from the defense industry and senior military leaders, creates a career path for computer scientists to prepare the military for the future AI battle.
In contrast to the Senate, the House AI Caucus, only put forth one major piece of AI legislation, the Fundamentally Understanding the Usability and Realistic Evolution of Artificial Intelligence Act of 2017. This legislation “establishes a federal advisory committee to examine and wrestle with the economic opportunities and impacts emerging AI technologies will have in many aspects of American life.” In line with this legislation, the Commerce Department becomes responsible for leading government policy on AI and submitting recommendations on how government and business can work together to advance this capability.
Not waiting to react to a national emergency circa 1947, 1986, or 2004, the USG thankfully appears to be learning from the past. It has taken a truly whole of government approach to AI during the Trump Administration, enlisting amongst others DoD, DHS, and Commerce, as well as Congress, to position the U.S. to capitalize on the benefits offered by this technology. To capitalize on this first mover advantage, the USG must continue to work with industry and academia to develop industry standards, developing strong, collaborative AI relationships across government, educate Americans on the benefits AI, as well as continuing to create career fields in government for computer scientists. The costs are too high, indeed they may be existential, for the USG to revert to its traditional reactive posture.
Dr. Mark Grzegorzewski is a Senior Defense Consultant.