AI Metrology Center

NIST’s AI Metrology Center is a set of resources within the AI Resource Center (AIRC) that supports the broad adoption of trustworthy AI. The Center integrates metrics, methodologies and tools with the NIST AI RMF trustworthy characteristics and lifecycle stages, enabling organizations to identify and use the most appropriate methodologies for their AI use cases. Organizations are encouraged to select measurement approaches that can strengthen their own testing, evaluation, validation, and verification (TEVV) and governance processes, facilitating more reliable deployment and iteration for AI systems. Inclusion of resources in the Center does not constitute NIST endorsement, validation, or determination of suitability of any listed methodology, metric, or tool.

Community submissions

How submissions work

Community submissions of AI metrics and measurement methodologies are open. Proposals for inclusion in the Center are accepted and reviewed in the open on GitHub: a submission is a pull request that adds one YAML file describing a single metric or measurement method, review happens publicly on that pull request, and an accepted submission is considered for publication here.

The usnistgov/ai-metrology-submissions repository holds the submission guide, the authoritative format specification, and worked example submissions kept open so you can read a real review end to end.

Submit your metric for consideration

This is a new process and we expect to refine it as we go. Do not submit proprietary or confidential information — submission files and all review discussion are publicly visible.

AI Metrics and Measurement Methods

    Community feedback

    The AI Metrology Center is a living resource and is expected to evolve over time. Individuals are encouraged to provide feedback about the content of the Center by emailing aimetrologycenter@nist.gov; we particularly welcome feedback on the filter categories, such as "Primary TEVV Application" or "TEVV Instrument." AI Metrology Center content updates will be released periodically.

    Disclaimer

    This collaborative resource is hosted by NIST. Content reflects contributions from multiple parties and does not imply endorsement or agreement by NIST. References to commercial products, systems, or trade names are for identification only and do not constitute recommendation or preference. Information is provided "as is," without warranties of any kind, including accuracy, completeness, merchantability, fitness for a particular purpose, or non-infringement.