Toward Resilient Human-AI Collaboration: A Lifecycle Taxonomy of Sociotechnical Risks and Cascading Failures
This paper presents a lifecycle-oriented synthesis of human-AI collaboration risks spanning four stages: task allocation, interaction, feedback, and adoption, and proposes a conceptual interaction model that illustrates how these risks emerge from sociotechnical drivers, interact through cascading pathways, and ultimat...