The White House has released a plan for developing technical standards for artificial intelligence through the National Institute of Standards and Technology, following an AI initiative from a February executive order and efforts to keep the U.S. on top.
Titled “U.S. Leadership in AI: A Plan for Federal Engagement in Developing Technical Standards and Related Tools,” the plan opens with guidance for federal agencies, the private sector and academia to develop technical standards and tools to support trustworthy AI systems.
It explains why a plan is in place for federal engagement in AI standards, and the importance of doing so for the future of the U.S. economy and national security, and for the country to remain a leader in AI.
According to the February executive order, the plan aims to: “Ensure that technical standards . . . reflect Federal priorities for innovation, public trust, and public confidence in systems that use AI technologies; and develop international standards to promote and protect those priorities.”
The plan claims the country’s leadership in AI will benefit from federal engagement in AI standards-related efforts in areas of supporting and conducting AI research, actively engaging in AI standards development, procuring and deploying standards-based products, and implementing the necessary regulatory policies where needed.
“Public trust, security, and privacy considerations remain critical components of our approach to setting AI technical standards,” said Michael Kratsios, federal chief technology officer. “As put forward by NIST, Federal guidance for AI standards development will support reliable, robust and trustworthy systems and ensure AI is created and applied for the benefit of the American people.”
And according to the plan, certain AI technical and non-technical standards are needed – including safety, accuracy, usability, interoperability, security, reliability, data and ethics; and governance, privacy, and societal and ethical considerations.
Certain AI standards-related tools are needed, too, like datasets and metadata in standardized formats, tools for capturing and representing knowledge and reasoning in AI systems, and fully documented use cases that provide information about specific applications of AI technologies.
Other tools include benchmarks and evaluations, testing methodologies, metrics to measure and characterize AI technologies, AI testbeds, and tools for accountability and auditing.
The plan also identifies which standards development efforts call for federal engagement, and which efforts agencies should prioritize. For example, agencies should focus on consensus-based efforts, meaning decision-making is based on established agreements understood by al involved parties, and decisions are reached on general agreement.
Others are inclusive and accessible efforts, relating to input from diverse and balanced communities of users, developers, vendors and experts; and result in globally relevant and non-discriminatory standards, meaning standards that avoid locking in technologies or becoming non-tariff trade barriers.
And certain AI standards characteristics deserve priority as well, including innovation-oriented standards, cross-sector applicability, those with clearly stated scope and intended use, regularly updated, human-centered, those with an understanding of AI technology risks and implications, and sensitive to ethical conditions.
The plan also expands on the necessary levels of federal engagement in AI standards from monitoring to leading, and the practical steps agencies can take to engage. To close, the plan recommends agencies start committing to more consistent and long-term engagement in AI standards development activities to speed the pace of AI technology development.