Washington, D.C. — TechNet, the national, bipartisan network of tech CEOs and senior executives, submitted comments from President and CEO Linda Moore on the National Institute of Standards and Technology’s (NIST) initial public draft of “Guidance and Templates for Public-Facing AI Documentation,” proposed voluntary standards for artificial intelligence (AI) developers to explain their AI model capabilities, data characteristics, and risk factors in an accessible manner. Moore welcomed this guidance as an important step to demystify AI and support effective risk management, while making a number of recommendations to strengthen its utility as an internationally applicable standards model.
“TechNet welcomes this effort from NIST to promote a shared, common-sense understanding of artificial intelligence. This guidance has the potential to serve as the basis for a universal AI ‘instruction manual’ — using common language to explain how AI models are built, how they work, and what they can do. With modest improvements to make it more practical and applicable globally, we believe this guidance can play a fundamental role in supporting innovation and fostering greater trust in responsible AI development,” said Linda Moore, TechNet President and CEO.
Specifically, TechNet recommended that NIST make improvements to the draft guidance in four specific areas:
- Promote global harmonization and interoperability with existing international standards;
- Explicitly recognize shared responsibility in risk management across the AI lifecycle according to each actor’s role and practical ability to identify, manage, and mitigate the relevant risk;
- Preserve flexibility for organizations to use automated, human, or hybrid approaches to monitoring, evaluation, and risk mitigation; and
- Facilitating the reuse of existing documentation to minimize duplicative burdens.