NIST AI News & Updates

NIST AI Standards Development and Global Engagement

NIST has released its final 'A Plan for Global Engagement on AI Standards' following public comment. This plan outlines NIST's strategy for promoting and developing AI standards globally to foster innovation and public trust in AI systems.

NIST AI Report Expands Evaluation Toolbox with Statistical Models

NIST has released a new publication, NIST AI 800-3, which introduces statistical models to improve the validity and robustness of AI system evaluations. The report distinguishes between benchmark accuracy and generalized accuracy and proposes the use of generalized linear mixed models (GLMMs) for more precise AI performance measurement.

NIST Report on AI Monitoring Challenges

NIST has released a new report, NIST AI 800-4, detailing challenges in monitoring deployed artificial intelligence systems. The report identifies six common categories of monitoring and highlights gaps and barriers to effective AI system oversight, based on practitioner workshops and literature reviews.

NIST and GSA MOU to Boost AI Evaluation Science in Federal Procurement

NIST's Center for AI Standards and Innovation (CAISI) has signed a Memorandum of Understanding with the General Services Administration (GSA) to enhance AI evaluation science for federal procurement. This collaboration aims to support the USAi platform by developing robust methodologies for assessing AI model performance and security.