CTRL+Story / Developing Configuration Literacy for Human-AI Collaboration in XR

Abstract

This doctoral colloquium paper presents work in progress investigating how an interdisciplinary team develops the capacity to collaborate with a multi-agent artificial intelligence (AI) system in a virtual reality (VR) laboratory. CTRL+Story brings together a filmmaker, designer, computer scientist, dancer and educator working alongside three distinct large language model (LLM) agents (two configured as creative participants, one as a reflexive Assessor) to co-create an immersive narrative experience. Existing frameworks for artificial intelligence in education (AIEd) position systems as tools acting on individual learners in bounded interactions. This research addresses what current frameworks cannot account for: how groups develop configuration literacy, defined as the capacity to configure, critically evaluate and iteratively refine persistent AI systems as collaborative participants. Informed by Karen Barad’s agential realism and Silvia Rivera Cusicanqui’s ch’ixi, the research treats the laboratory apparatus (collaborative participants, VR environment, facilitation structure and documentation infrastructure) as the primary research instrument, examining how distributed authorship and traceable attribution emerge through collaborative configuration. The research employs a longitudinal design, generating primary data through session transcripts, configuration logs and Assessor analysis, with the XR artefact serving as a material trace of configuration literacy as it develops within the CTRL+Story lab.