User Science for AI is an emerging community. This vision is designed to serve as a starting point to foster discussion and collaboration on the topic. At the same time, it serves to distinguish this community from other adjacent communities such as Human-Centered AI and Explainable AI. For instance, we are not focused on developing new foundational models, improving benchmark accuracy, or AI safety/governance as primary topics. We welcome feedback and contributions from members in the future to help shape its future direction.
Five key pillars support the initial research vision of this community:
Develop a cumulative scientific understanding of how people perceive, interact with, and collaborate with AI-enabled systems.
Advance the methodologies, measurement instruments, and evaluation practices needed to study AI-enabled experiences with scientific rigor.
Bring together researchers from HCI, Human-AI Interaction, AI, psychology, human factors, design, cognitive science, and related disciplines to address shared methodological challenges.
Promote theories, open resources, and shared research practices that allow knowledge about AI-enabled experiences to accumulate across studies, systems, and domains.
Ensure that advances in AI are accompanied by equally strong advances in understanding and evaluating the experiences of the people who use these technologies.