Within-person reciprocal processes between AI feedback literacy and agentic engagement: a three-wave random-intercept cross-lagged panel study

Artificial intelligence (AI) feedback has become a pervasive feature of university learning, yet whether students' capacity to engage productively with such feedback co-develops with their broader learner agency remains poorly understood. Drawing on student feedback literacy theory, self-determination theory as a supplementary motivational lens, and the mutualism principle, this study tested whether AI feedback literacy and agentic engagement reciprocally predict each other over time at the within-person level and exploratorily examined whether they share a stable trait-level association. A three-wave panel study of 480 Chinese undergraduates (N = 399 with complete data across all waves), with measurements at six-week intervals across one semester, was analyzed using a random-intercept cross-lagged panel model (RI-CLPM). At the between-person level, students who scored systematically higher in AI feedback literacy across the semester also scored systematically higher in agentic engagement, with a moderate and statistically significant trait-level correlation between the two random intercepts (r = 0.315, p < 0.001), supporting Hypothesis 1. At the within-person level, both cross-lagged pathways were positive and significant: deviations in AI feedback literacy predicted later deviations in agentic engagement (β = 0.222, p < 0.001), and the reverse pathway was likewise significant (β = 0.263, p < 0.001). A Wald test indicated that the two reciprocal pathways did not differ significantly in magnitude (ΔB = 0.002, p = 0.972), consistent with a symmetric mutualism. A model comparison confirmed that the RI-CLPM provided substantially better fit than a traditional cross-lagged panel model, with the RI-CLPM revealing substantially larger within-person reciprocal effects once stable trait-level covariation was properly isolated. The findings extend the mutualism principle from a coupling of person-internal constructs to a coupling between a person-internal competency and a person-environment behavioral interface. They also raise the possibility, in a Confucian Heritage Culture context where agentic engagement has historically been culturally muted, that AI feedback may offer a less culturally-loaded venue in which students can exercise agency. The study repositions AI feedback literacy from a static skill into a dynamic developmental resource that is constituted, in part, by the very engagement it enables.