When AI Agents Become Immersive:A Human-Centered Design and Evaluation Framework

Abstract

Immersive learning is often associated with advanced interfaces or emerging technologies. In educational contexts, however, immersion is shaped by how learners experience meaning, agency, and relational presence within a learning environment. As educational AI agents become more common, a central design challenge emerges: when do AI agents genuinely support immersive learning, and how can this be intentionally designed? This iLEAD paper presents a human-centered approach, positioning immersion as an emergent learning outcome rather than a standalone feature. It introduces a Human-Centered AI App Design Starter Guide that helps educators and designers clarify learning purpose, define human and AI roles, and design agent interactions that support reflection, creation, and application. The session also introduces a complementary Human-Centered Evaluation Tool for Educational AI Agents that examines whether design intent translates into an immersive learner experience across dimensions such as relational presence, learner agency, engagement, contextual coherence, and reflective depth. Together, these tools offer a practical framework for designing and evaluating educational AI agents that foster meaningful immersive learning.