On August 9, 2019, the National Institute of Standards and Technology (NIST) released “U.S. Leadership in AI: A Plan for Federal Engagement in Developing Technical Standards and Related Tools” (the Plan) in response to Executive Order 13859 (EO), as reported on here. In accordance with the EO, the Plan outlines the following priorities for federal engagement: (1) ensure technical standards minimize vulnerability to attacks from malicious actors; (2) reflect federal priorities for innovation, public trust, and public confidence in systems that use artificial intelligence (AI) technologies; and (3) develop international standards to promote and protect those priorities. With emphasis from both public and private sector, NIST calls for flexible AI standards in regulatory and procurement actions, as well as the prioritization of multidisciplinary research and expansive public-private partnerships. Based on the Plan, companies are likely to see an increased number of opportunities to participate and assist the Federal Government in the standard development process, while simultaneously being put on notice that standards in this burgeoning industry may be forthcoming.

The Importance of U.S. Involvement in Standards and Artificial Intelligence

The United States’ global leadership in AI is dependent upon the Federal Government’s active participation in AI standards development. To maintain this leadership, NIST calls on the active participation of both the private sector and academia. Currently, AI standards are either cross-sector (e.g., applications and industries) or sector-specific (e.g., healthcare and transportation). But, NIST emphasizes that the Federal Government should foster collaboration between the two camps where there may be several existing cross-sector or sector-specific technology standards applicable to AI that originally were developed for other technologies, e.g., cybersecurity and privacy. On this point, NIST urges the consideration of these already existing standards before the creation of new ones.

In that vein, NIST requests a consistent set of “rules of the road” for AI. NIST believes these will enable market competition, preclude barriers to trade, and allow innovation to flourish. They also will ensure that AI technologies and systems meet critical objectives for functionality, interoperability, and trustworthiness. Yet, in pursuit of these objectives, the standards must include societal and ethical considerations in IT, as well as aspects of trustworthiness (e.g., explainability and security).

NIST also recognizes the importance of establishing aspirational principles and goals to address legal, ethical, and societal issues in developing AI standards. Here, NIST notes that there has been a “first step” toward standardization of these principles when the Organization for Economic Co-operation and Development Council’s (OECD) Recommendations on AI (Recommendation), as reported here, emphasized the need for the same considerations in the promotion of AI. Thus, NIST calls on the Federal Government to ensure the same cooperation and coordination across federal agencies and private sector stakeholders and to continue its engagement in international dialogues on AI standards. However, while the United States endorsed the Recommendation and adopted these principles, and there is “broad agreement” that these considerations must factor into AI standards, NIST cautions that the current path forward to factor in these issues is unclear.

NIST Urges the Federal Government to Engage in the Creation of AI Standard Priorities

To maintain a leadership position in AI, the Federal Government must create a “purpose-driven role in AI standards development.” To that end, NIST addresses priorities for Federal involvement, the different level of U.S. engagement, and practical first steps. For involvement, the Plan calls for prioritizing AI standards efforts that were “inclusive and accessible, open and transparent, consensus-based, globally relevant and non-discriminatory.”

NIST also provides recommendations for the type of AI standards that deserve priority consideration by the Federal Government. For example, NIST recommends focusing on standards that are innovation-oriented, are applicable across multiple sectors, are effective in monitoring and evaluating AI system performance, and are sensitive to ethical considerations, to name a few.

NIST urges Federal Government involvement in the standard development process to not only protect U.S. dominance in AI, but also to ensure that U.S.-based companies are not excluded or disadvantaged as standards are developed. NIST suggests four different levels of involvement that range from passive to more interactive. The Plan explains that the Government can monitor, participate, influence, and lead. NIST notes that engaging in any of the four levels of involvement requires qualified and competent Federal Government participants, including Federal employees and contractors, to assist with the standards development process.

NIST also provides suggestions on the first practical steps agencies can take to begin engaging in the AI standards process. These include identifying the AI technologies that can be used to further the agency’s mission; conducting a gap analysis to determine if there are already standards in place that can be used or if standards need to be developed; coordinating with agencies with similar needs; and identifying, training, and enabling staff to participate in standards development.

NIST Provides Recommendations for Advancing U.S. Leadership in AI

NIST also calls for the Federal Government to advance its role in AI leadership to help “speed the pace of reliable, robust, and trustworthy AI technology development.” NIST provides the following four recommendations to the Federal Government in the advancement of AI standards:

  1. Agencies should share information, leadership, and coordination regarding standards development. Growing a cadre of Federal staff that encourages the participation and expertise in AI standards and standards development would be particularly useful.
  2. Agencies should participate in focused research on how to incorporate “trustworthiness” into AI standards and tools.
  3. Agencies should use public-private partnerships to develop AI standards and tools.
  4. For U.S. economic and national security purposes, the Federal Government should engage strategic international partners in the development of AI standards.

Significance of NIST’s Plan

The Plan from NIST is significant in that it provides meaningful guidance, recommendations, and concrete steps for the Federal Government to use in developing a set of AI Standards. But, as NIST emphasizes throughout the Plan, the Federal Government must engage and involve private industry to formulate these AI Standards. For example, NIST urges agencies to study and understand the approaches technology companies are taking to steer their own AI development efforts. Likewise, NIST calls for increased investment in research that focuses on understanding AI trustworthiness and incorporating those metrics into future standards, where private industries will likely have a key role to play. The expansion of public-private partnerships is integral to help inform federal AI standards. Therefore, companies should expect to see an increased number of opportunities to participate and assist the Federal Government in the standard development process. Meanwhile, however, companies active in or considering entering into the AI industry should consider engaging in – or at least monitoring – the standard-setting process that the Plan signals is about to get underway.