Exploring the Mediating Role of Perception of AI-Driven Personalization and Green Purchase Intention in the Relationship Between Environmental Knowledge, Environmental Attitude, and Green Purchasing Behavior Among Youth in Java Island

Sustainable consumption is a global imperative, yet its behavioral drivers vary across cultural and technological contexts. This study examines how environmental knowledge (EK), environmental attitude (EAT), and perception of AI-driven personal-ization (PAI) shape green purchasing intention (GPI) and green purchasing behavior (GPB) among youth in Java, Indonesia. A cross-sectional survey of n = 517 university students (17–25 years) was analyzed using PLS-SEM (SmartPLS 4.0). The extended TPB model showed adequate measurement quality (e.g., SRMR = 0.074; HTMT < 0.90) and explanatory power (R²: EAT = 0.436; GPI = 0.373; GPB = 0.553; PAI = 0.077). Key structural paths were significant: EK → EAT (β = 0.661, p < 0.001); EAT → GPI (β = 0.445, p < 0.001); EAT → GPB (β = 0.366, p < 0.001); GPI → GPB (β = 0.444, p < 0.001); PAI → GPI (β = 0.136, p = 0.001), while PAI → GPB was not. Mediation tests (5,000 bootstraps) confirmed ro-bust indirect effects via EAT and GPI (e.g., EK → EAT → GPB; EAT → GPI → GPB; EK → EAT → GPI → GPB). Predictive assessment indicated Q² > 0 for endogenous constructs (EAT = 0.228; GPI = 0.220; GPB = 0.337), and PLSpredict favored the PLS-SEM model over a linear benchmark on 26 of 40 RMSE/MAE comparisons, evidencing out-of-sample utility. The findings suggest that strengthening knowledge to cultivate supportive at-titudes, coupled with transparent, value-aligned AI personalization, can elevate inten-tion and translate into greener purchases. Implications are offered for platform design, education, and policy in emerging markets.