Human–Twin Interaction (HTI)

Digital twins have matured from static simulation artifacts into continuously evolving, data-driven counterparts of physical systems. Yet, a review of current literature reveals that the field remains profoundly system-centric, emphasizing fidelity, prediction accuracy, and automation, while under-theorizing the role of the human. Although high-level concepts like human-centricity and explainability are widely discussed, the technical approaches to implementing them—such as ontological foundations, unified models, human-environment interaction, and federated semi-supervised learning—remain siloed and address interaction challenges in isolation. This position paper introduces Human–Twin Interaction (HTI) as a unifying, interaction-focused framework for understanding and designing human-centered digital twin systems. HTI is proposed as an umbrella construct that consolidates these fragmented technical efforts under a single, coherent interaction paradigm. We argue that training-oriented digital twins provide a revealing context where the limitations of current paradigms become visible. The paper articulates the conceptual foundations of HTI, situates it within a history of human-system failures and current digital twin research, proposes a conceptual framework, and illustrates its application through a training-oriented case. We conclude by outlining research directions for developing digital twin systems that do not merely optimize performance, but actively support human learning, judgment, and responsibility.