A new framework for measuring how well humans rely on set-valued AI advice has been proposed by researchers. Set-valued advice, which provides a range of possible outcomes rather than a single prediction, is becoming increasingly used in human-AI collaboration to communicate uncertainty and improve decision-making. The framework, developed for both classification and regression tasks, includes metrics such as correct reliance rate on AI and quality of AI reliance, which capture nuances in human-AI interaction that existing measures miss. The authors claim their work addresses a gap in current research, which has focused on point predictions rather than set-valued advice. --- Why it matters: This matters to researchers because it provides a formal framework for evaluating the effectiveness of set-valued AI advice, which is becoming increasingly important as AI systems are used to support human decision-making. By understanding how humans rely on this type of advice, developers can improve the design and implementation of AI systems that provide more accurate and useful recommendations. Source: https://arxiv.org/abs/2606.06081