<![CDATA[This study addresses the critical gap between the widespread adoption of AI-powered educational technologies and the persistently low life science achievement among Indonesian elementary students. The objective is to uncover the mechanisms and boundary conditions shaping this relationship by testing a moderated mediation model in which self-regulated learning mediates the effect of perceived AI-supported learning on life science achievement, and prior life science knowledge serves as a critical moderator. A quantitative cross-sectional survey was conducted with 253 elementary students. Using partial least squares structural equation modeling, the findings reveal that self-regulated learning fully mediates the relationship between perceived AI-supported learning and life science achievement (β = 0.134, p < 0.001), while the direct effect remains non-significant (β = −0.058, p = 0.361). Prior life science knowledge significantly moderates both pathways (β = 0.172, p = 0.004; β = 0.158, p = 0.004). AI tools enhance achievement primarily by fostering self-regulatory capacities, with significant benefits restricted to students whose prior life science knowledge exceeds 1.71 standard deviations above the mean. These results advance a conditional, mediated theoretical framework for technology-enhanced life science learning and highlight the practical necessity of differentiated AI implementation paired with explicit metacognitive instruction.]]>
Copyrights © 2026
