NIST Launches New ARIA Program to Assess Societal Risks and Impacts of AI
The National Institute of Standards and Technology (NIST) has introduced a pioneering program, Assessing Risks and Impacts of AI (ARIA), designed to evaluate the societal risks and impacts of artificial intelligence systems. ARIA will focus on understanding the real-world implications of regular interactions between people and AI, providing a comprehensive view of how AI functions within societal contexts once deployed.
The primary goal of the ARIA program is to develop methodologies and metrics that quantify AI systems’ performance and their broader societal impacts. This initiative is crucial to the U.S. AI Safety Institute’s efforts to establish the foundation for trustworthy AI systems.
ARIA aligns with recent NIST announcements, marking the 180-day milestone of the Executive Order on trustworthy AI and the unveiling of the U.S. AI Safety Institute’s strategic vision and international safety network.
Ensuring Safe and Trustworthy AI
Understanding AI’s societal impacts is critical for mitigating risks and maximizing benefits. The ARIA program is designed to meet real-world needs as AI technology usage grows, supporting the U.S. AI Safety Institute and expanding NIST’s engagement with the research community. By establishing reliable methods for testing and evaluating AI’s functionality in real-world scenarios, ARIA aims to ensure AI systems are safe, secure, private, and fair once deployed.
Building on the AI Risk Management Framework
ARIA builds on the AI Risk Management Framework released by NIST in January 2023. This framework recommends using both quantitative and qualitative techniques to analyze and monitor AI risk and impacts. ARIA will operationalize this risk measurement function by developing new methodologies and metrics for assessing AI systems’ safety and functionality within societal contexts.
The program will go beyond evaluating AI models in laboratory settings, considering how AI systems perform in realistic scenarios when used regularly by people. This holistic approach will provide a broader understanding of the net effects of AI technologies on society.
Supporting Safe, Secure, and Trustworthy AI Systems
The results of the ARIA program will support NIST’s collective efforts, including those of the U.S. AI Safety Institute, in building the foundation for safe, secure, and trustworthy AI systems. By offering a detailed understanding of AI’s societal impacts and developing reliable testing methods, ARIA will guide the responsible and beneficial integration of AI into society.
Stay tuned for more updates on AI, like the NIST ARIA program.