Redefining and measuring student agency in AI-assisted learning: Development and validation of the agentic engagement with AI (AE-AI) scale

Student agency has become a central concern in learning with AI. With growing capability and adaptiveness, GenAI tools are no longer merely passive tools but responsive participants in learning. This shift makes agency relational and reciprocal: students exercise agency through intentional engagement with AI, while AI responses shape how students express, adjust, and develop that agency. Existing instruments, however, continue to frame agency as student-centered control over external tools and environments, positing students as the primary initiator and regulator of learning. This misalignment leaves the relational and reciprocal nature of agency in AI-assisted learning unmeasured. To address this gap, the present study adopts the construct of agentic engagement and conceptualizes student agency in AI-assisted learning as a proactive , adaptable , and reciprocal process enacted through ongoing exchanges between human intentionality and machine adaptivity. To develop the Agentic Engagement with AI (AE-AI) Scale, an integrated inductive–deductive approach was used to generate an initial pool of 28 items from student interviews ( N = 26). The scale was refined and validated through expert review, exploratory factor analysis ( N = 340), confirmatory factor analysis ( N = 256), and criterion validity checking. The final 16-item scale comprises four factors: Adaptive Direction , Critical Integration , Cross-Source Inquiry , and Reflective Calibration . Together, these factors redefine student agency as a process of human-AI coagency grounded in adaptability, epistemic responsibility, and distributed inquiry. The study provides a validated instrument for observing student agency and opens new directions of examining the evolving synthesis of human and machine intelligence in education.