In February 2019, the President issued Executive Order 13859, which directed the Secretary of Commerce, through the National Institute of Standards and Technology (NIST), to issue “a plan for Federal engagement in the development of technical standards and related tools in support or reliable, robust, and trustworthy systems in AI technology.” NIST, with public and private sector input, issued the AI plan on August 9, 2019. The plan “provides guidance regarding the important characteristics of standards to help agencies in their decision making about AI standards.”
Federal Engagement Will Promote Development of AI Standards
The U.S. is the global leader in AI technology. NIST believes that getting the government involved in the development of AI standards will further U.S. leadership in AI. Use of standards facilitates technological advancement and market efficiency. AI standards that set forth specific guidelines will, NIST posits, ensure that AI technologies meet objectives for functionality, interoperability, and trustworthiness.
But NIST does not necessarily envision direct government involvement in the development of standards. In the U.S., the private sector typically develops standards with federal agencies contributing to and using these standards. The federal role includes sharing agency requirements with standards projects, providing technical expertise on standard development, incorporating standards into regulations, and citing standards in procurements.
NIST believes that federal contribution to and use of standards will help synthesize existing standards. Right now, AI standards are either (1) cross-sector, or horizonal, meaning that they apply across applications or industries; or (2) sector-specific, or vertical, which means they apply to specific industries like health care or transportation. NIST envisions the federal government playing a role in fostering collaboration between cross-sector and sector-specific developers.
How will the Government be Involved?
The NIST plan sets forth priorities for federal involvement. The government should prioritize AI standards that are consensus-based, inclusive and accessible, open and transparent, and globally relevant and non-discriminatory. The NIST plan also lists the type of characteristics the government should consider inusing AI standards, including whether the standards are innovative, applicable across sectors, regularly updated, effective in measuring system performance, and sensitive to ethical considerations (i.e., minimize bias and protect privacy).
NIST also discusses the various ways the government can get involved in the development of standards: (1) monitoring the development of standards, (2) participating in the development through comments, (3) influencing development through formal and informal discussions and by providing expertise, and (4) leading development by serving as a standards project editor or acting as the liaison between standards groups. Each of these levels of engagement will require qualified government participants—i.e., federal employees or government contractors.
The NIST plan further advises on some practical steps that agencies can take to begin engaging in standard development. For instance, agencies can identify technologies that can be sued to further the agency’s mission, know the existing policies and statutes that relate to use of standards, and use appropriate standards if they exist.
Finally, NIST makes more general recommendations for ways agencies can facilitate the development of reliable and robust AI technology:
- Shore up AI standards-related knowledge and coordinate on standards development.
- Promote focused research to explore how aspects of trustworthiness can be incorporated into standards and standards-related tools.
- Support and expand public-private partnerships in developing standards.
- Strategically engage with international parties to advance AI standards for U.S. economic and security needs.
The NIST plan provides guidance on how the government should engage with the development of AI standards, but we don’t yet know the exact steps agencies will take in this development. It is clear, however, that the federal government is encouraged to partner with the private sector, which means government contractors will likely participate in developing standards. Contractors will thus have an opportunity to define AI standards and determine how they are incorporated into regulations and solicitations going forward.