AI and the United States

In response to the 11 February 2019 US "Executive Order on Maintaining American Leadership in Artificial Intelligence," NIST is requesting information about technical standards and related tools for artificial intelligence during an open comment period ending 31 May. Comments may be submitted via email to ai_standards@nist.gov. More information can be found on the Federal Register.

In this article, SPIE/OSA Congressional Fellow Benjamin Isaacoff lays out the goals and gaps of the current US strategy for AI.

In February, US President Trump issued an executive order laying out the "American AI Initiative" (AAII). This executive order appears to be an aspirational start to a full-fledged US AI National Strategy. However, unlike the AI plans of peer and competing countries, the US AAII is not a detailed substantive plan that will, in and of itself, meaningfully move the nation forward on AI.

The last several years have seen countries across the world racing to get out in front of AI technologies and ensure that their companies, researchers, and/or militaries have the strategic advantage and aren't left behind in the (presumably) forthcoming AI revolution. No two countries have identical AI national strategies, but most feature large investments in AI R&D, efforts to boost a computer science-literate workforce balanced with policies aimed at helping workers displaced by automation, and calls to increase public and private sector uptake of AI-based tools.

The substance of the AAII covers most of the important areas that a comprehensive AI National Strategy should. The AAII covers AI R&D by directing federal research funders to prioritize funding AI R&D. The critical issues of data availability and computing resources are similarly addressed, though somewhat broadly, by directing federal agencies to work on these areas while balancing countervailing privacy concerns. The AAII also mentions the often-overlooked issue of technical standards development for AI tools, in this case directing the National Institute of Standards and Technology (NIST) to issue a plan. The workforce section of the AAII focuses on traditional education support, though it does not delve into worker retraining or any "future of work" considerations. Finally, the AAII calls for balancing the need to keep international markets open for US AI industries, while still protecting and prioritizing US companies.

However, the AAII executive order omits some information as well. For example, it directs various federal agencies to make and implement their own plans to achieve these goals. And, although the AAII calls for investments in AI R&D, it doesn't specify that Congress should fund these initiatives. Instead, much of the funding language suggests that R&D funders should reallocate existing funds, as opposed to growing the overall R&D investment. The AAII also doesn't provide detail about the R&D investments the White House is calling for, an approach in contrast to many foreign AI National Strategies that are more strategic about the types of AI R&D they fund. Finally, the AAII does not mention directing public sector uptake of AI-based tools, something that AI policy experts and foreign governments believe to be a win-win approach to spurring AI development and deployment while increasing good governance.

Executive orders are often light on the details, and this usually isn't a problem when they represent just one piece of an overall policy agenda. However, the AAII has left many observers unclear about White House priorities. For instance, the AAII calls for increasing AI R&D, however White House budget requests submitted to Congress have included large cuts to federal R&D funding. The US Congress has thus far chosen to increase R&D funding for the last several years, despite the budget requests to decrease it.

The AAII is also very interested in promoting and protecting the United States' current leading position in AI relative to other countries. One of the most effective strategies to ensure US leadership—one that has been wholeheartedly employed for the last century—is to attract the best and the brightest from around the world to study and work in the US. However, current immigration policies are making it more difficult for immigrants to come to the US for work or study. Excluding foreign researchers and entrepreneurs will likely harm US prospects and grow those of other countries.

The many possible paths forward

Congressional interest in a wide variety of AI policy topics has grown steadily. This independent congressional interest combined with the AAII out of the White House may lead to numerous meaningful individual AI policies being enacted. For example, it is likely that congressional appropriators will continue to generously fund federal R&D on AI. Alternatively, it's possible that Congress and the White House could come together to develop an actual full-fledged AI National Strategy for the US.

With the newly formed Senate AI Caucus, and the already established House AI Caucus, Congress is organizing itself to write and pass meaningful AI legislation. Many of these bills address shortcomings of the AAII. The FUTURE of AI Act, for example, seeks to create an advisory committee in the Department of Commerce that would effectively develop recommendations for an AI National Strategy. The AI in Government Act is focused on promoting public sector use of AI tools. The GrAITR (Growing Artificial Intelligence Through Research) Act is focused on funding AI R&D and is both generous with the amount funding and is very specific about where the funding would go. The AI Jobs Act of 2019 directs the Department of Labor to study how the rise of AI will affect the US workforce. H.Res.153 is a resolution supporting the development of guidelines for ethical issues in AI.

At this point it's unclear what the future will hold for federal AI policy in the US. Will Congress lead on AI policymaking, and if so, will it be with the current approach of a bevy of individual policies focused on specific issues, or will Congress try and put together something more comprehensive? Or, possibly, is the AAII truly an aspirational start to a government-wide effort to craft and implement a full-fledged AI National Strategy? What we do know now is that it's clear that everybody—the White House, US Congress, and foreign governments—recognizes the importance of AI policy. The only question is what they're going to do about it.

Benjamin Isaacoff is the 2018-2019 OSA & SPIE Arthur H. Guenther/AAAS Congressional Science and Engineering Fellow.

The SPIE/OSA Arthur H. Guenther Congressional Fellow works in the office of a U.S. Senator or Representative or with a Congressional Committee to gain first-hand knowledge of congressional operations, contribute to the policy-making process, and forge links between the engineering, scientific, and public policy communities. Fellows also gain a perspective that will enhance industrial, academic, and government careers. The Fellowship is an ideal way to spend an academic sabbatical or leave of absence from a company.

SPIE co-sponsors a Congressional Fellow with The Optical Society each year. Fellows receive a salary, as well as a stipend for health insurance, travel, and relocation expenses. Applications for the 2020-2021 Fellowship Program will be due at the beginning of 2020.