From Framework to Evidence: Testing Explainable AI Feedback for Leadership Learning in Collaborative VR under the C²L-AI Model

This study advances a testable account of human-AI collaborative competence via the C²L-AI framework. We integrate Activity Theory, Distributed Cognition, and sociomateriality to reconceptualize collaboration, communication, and leadership alongside AI Interaction Competence. We operationalize constructs with an Evidence-Centered Design (ECD) multimodal matrix spanning NLP, social/epistemic network analysis, and VR behavioral analytics. To generate causal evidence, we propose a three-arm randomized controlled trial in a multi-user VR leadership task comparing Explainable AI (XAI) feedback, standard feedback, and no feedback. We hypothesize that XAI yields greater gains in leadership, communication, and AI interaction competence, mediated by improvements in team cognitive architecture (shared mental models, transactive memory). The work offers a unified theory, measurable indicators, and an empirical pathway for designing effective, ethical human-AI learning systems.