AI-driven gamification and inclusive learning outcomes in higher education institutions: a structural equation modeling approach

Abstract

The integration of artificial intelligence (AI) in higher education has provided the opportunity to bring new possibilities into the realm of adaptive and data-driven learning environments. The aim of this study is to explore the impact of AI gamification on inclusive learning outcomes in higher education and the role of learner engagement, learning motivation, and personalized learning experience in this relationship. The data were collected using a quantitative, cross-sectional survey design from 303 students from higher education institutions in India, who reported prior experience of AI-based or gamified learning platforms. IBM AMOS was used for the Structural Equation Modeling (SEM). AIG had significant positive impacts on learner engagement ( β = 0.572, p < 0.001), learning motivation ( β = 0.483, p < 0.001), and personalized learning experience ( β = 0.566, p < 0.001). Inclusive learning outcomes were significantly predicted by all three mediating variables. Learning motivation emerged as the strongest predictor ( β = 0.466, p < 0.001), followed by personalized learning experience ( β = 0.455, p < 0.001) and learner engagement ( β = 0.388, p < 0.001). A significant negative direct effect was found between AIG and inclusive learning outcomes ( β = −0.256, p < 0.001), indicating a suppression effect: the net positive effect of AIG on inclusive learning outcomes is entirely mediated through the three psychological and experiential pathways. Bootstrapped mediation analysis (2,000 resamples) confirmed significant partial mediation via learning motivation ( β indirect = 0.199, 95% CI [0.132–0.268]), personalized learning experience ( β indirect = 0.198, 95% CI [0.129–0.261]), and learner engagement ( β indirect = 0.176, 95% CI [0.115–0.241]). Prior online learning experience had a significant positive effect ( β = 0.105, p = 0.004), and digital literacy had a significant positive effect ( β = 0.138, p < 0.001) on inclusive learning outcomes. Model fit was acceptable: χ 2 /df = 2.384, RMSEA = 0.069, CFI = 0.896, TLI = 0.883, IFI = 0.897. Incremental fit indices are marginally below the conventional 0.90 threshold, though χ 2 /df and RMSEA meet accepted benchmarks; effect sizes should be interpreted with appropriate caution. This study extends Self-Determination Theory by operationalizing AI-driven gamification (AIG) as a single adaptive construct within a SEM mediation model, revealing a suppression effect in which AIG promotes inclusive learning outcomes exclusively through engagement, motivation, and personalization pathways. The findings offer actionable guidance for platform designers, educators, and policymakers committed to building equitable AI-enhanced learning environments in higher education.