Guidelines for AI-supported virtual laboratories in science museums

Abstract Science museums are increasingly positioned as spaces for non-formal sustainability education, yet systematic frameworks for integrating AI-supported virtual laboratories into these environments remain scarce. This review article proposes a preliminary conceptual framework for the design of AI-integrated virtual laboratories in science museums, with a focus on citizen science participation and socioecological challenges. Through a Delphi-oriented, internal designer-led consensus process combining two rounds of multidisciplinary reflection with LLM-assisted synthesis, the study identifies thirteen dimensions: contextual understanding, operational sustainability, research support, educational design, audience personalization, technological infrastructure, physical-digital interaction, operability, ethics, participation, accessibility, interculturality, and international collaboration. Each dimension is characterized by possible AI capabilities and intended outcomes. Developed in connection with LúmiaLab, a virtual interdisciplinary laboratory currently in its conceptual design phase with the Museum of Environmental Sciences in Guadalajara, Mexico, the framework is contextually grounded but has not yet been independently or empirically validated. It contributes a structured theoretical-practical reference for subsequent co-design, piloting, and external validation rather than evidence of educational effectiveness or generalizability.