NIST releases draft AI standards plan
The National Institute of Science and Technology has released a draft of its plan for developing federal artificial intelligence standards, which could govern a range of technologies relating to AI, including biometric facial recognition.
NIST published its draft of “U.S. Leadership in AI: A Plan for Federal Engagement in Developing Technical Standards and Related Tools” for public comment, which it will accept up until July 19. The final version will be delivered to the White House on August 10, VentureBeat reports, to satisfy an executive order mandating the creation of a federal AI engagement plan within 180 days. The plan involves including a diverse community of stakeholders, such as users, developers, vendors, and experts from technical disciplines and non technical disciplines like ethics, economics, law, and public policy.
The report surveys the landscape of existing standards, and those in development by bodies like the IEEE and Defense Advanced Research Projects Agency (DARPA), and suggests that parallel efforts in other countries be tracked as well. ISO/IEC JTC 1 SC 37 is specifically identified in the appendix on existing standards efforts.
It then calls for a standards coordinator position to be created by the National Science and Technology Council (NSTC) Machine Learning/Artificial Intelligence (ML/AI) Subcommittee, for research which could increase trust in AI to be identified, and for auditing tools to be created to test AI models, datasets, and test environments.
The U.S. also recently updated its research and development strategy for AI, though some pundits have argued that the U.S. needs to avoid hindering its own efforts to keep up in the AI industry in other ways, such as with antitrust actions.