AI Recommendation Systems as Eco-Conscious Promotional Agents: Underlying Motivations for Sustainable Consumption in E-Commerce

AI recommendation systems are reshaping sustainable consumption in digital marketplaces, yet the psychological mechanisms governing consumer adoption remain underexplored in emerging economies. This study investigates consumer intention to adopt AI recommendation systems for sustainable consumption in Vietnam, drawing on an integrated framework combining the Information Systems Success (ISS) model, human-computer interaction (HCI) principles, UTAUT2, and Cognitive Dissonance Theory (CDT). Algorithmic trust and affective attitude are positioned as dual mediating pathways, with the AI-Environment Paradox incorporated as a formal moderating boundary condition. A structured survey of 403 e-commerce users was analyzed using PLS-SEM. System quality emerges as the strongest predictor of cognitive trust, while personalization exerts the greatest influence on attitude. Attitude is the dominant driver of behavioral intention, substantially outperforming trust and social influence. Performance expectancy and facilitating conditions yield negligible effects, reflecting the post-normalization of AI usage among digitally native consumers. Notably, the AI-Environment Paradox operates asymmetrically, attenuating the trust-to-intention pathway while amplifying the attitude-to-intention link through compensatory pro-environmental reasoning, a pathway-specific moderation effect that extends CDT into the domain of AI-assisted sustainable consumption. These findings reframe adoption logic for sustainable AI commerce and carry actionable implications for platform design, personalization strategy, and ecological communication in high-penetration digital economies.