A User Model to Enable Hybrid Human-AI Collaboration

The shift to Industry 5.0 is based on human-centric, resilient, and sustainable industrial systems, where AI is no longer a simple tool nor a replacement for human faculties, but rather a collaborator. Thus, we move away from a model of full automation in which the human is left out of the loop, to a human-centric vision in which the human is assisted, augmented, and catered to by AI-driven agents that proactively adapt to human needs. To achieve this vision, it is fundamental that AI agents understand human needs, intentions, and workflows in real-time, enabling more intuitive, adaptive and meaningful interactions. This paper introduces the socio-technical context stream, a framework that provides AI with a real-time, human-centered semantic understanding of industrial contexts. By continuously integrating human intent, workflows, and situational factors, AI agents can dynamically adapt their interactions, fostering deeper, more intuitive and natural Human-AI collaboration. This paper focuses on the definition of a context model, and how it can be populated to then feed AI-driven agents with meaningful contextual information, with a particular focus on modeling human actors and envisioning the interaction with Digital Twins.