NIST: AI Use Taxonomy - A Human-Centered Approach

• The "AI Use Taxonomy: A Human-Centered Approach" by the National Institute of Standards and Technology (NIST) provides a comprehensive framework for understanding and classifying AI systems' roles in human-AI interactions. It emphasizes the importance of focusing on human goals and outcomes to ensure AI contributes positively to these interactions.
• The taxonomy identifies 16 AI use activities, independent of specific AI techniques or domains, making it adaptable across various fields and technologies. These activities include content creation, synthesis, decision making, detection, digital assistance, discovery, image analysis, information retrieval/search, monitoring, performance improvement, personalization, prediction, process automation, recommendation, robotic automation, and vehicular automation.
• The paper outlines the taxonomy's purpose to aid in the measurement and evaluation of AI systems, improving their usability and trustworthiness characteristics as per the NIST AI Risk Management Framework (RMF). It serves a wide audience involved in AI design, deployment, and evaluation, providing a common language for describing human-AI activities. The taxonomy also highlights the need for future research in applying this classification to understand measurement challenges and to enhance AI system evaluation methodologies. Additionally, it presents a focus on usability, detailing the importance of assessing AI systems in terms of effectiveness, efficiency, and user satisfaction.

Original source: AI Use Taxonomy: A Human-Centered Approach | NIST