Can generative artificial intelligence drive sustainable behavior? A consumer-adoption model for AI-driven sustainability recommendations

Generative AI (GAI) has the potential to promote sustainable behavior through personalized recommendations; yet its effectiveness hinges on user trust—an issue that remains under-explored in the literature. Existing studies often focus on specific domains without addressing broader trust-building mechanisms or the cognitive and motivational factors needed for sustained engagement. This study investigates how trust shapes the adoption of GAI-driven sustainability recommendations by integrating the Elaboration Likelihood Model (ELM) and Expectancy-Value Theory (EVT) into a single framework. Using data from sustainability-oriented users, we examine how central route constructs-perceived information quality and utility-peripheral route constructs-anthropomorphism and interaction quality-enhance trust, while perceived information complexity and perceived risk moderate these relationships. Our findings indicate that high-quality, useful information enhances trust through cognitive engagement, whereas anthropomorphic design and interaction quality reinforce trust via the heuristic route. However, excessive complexity and privacy concerns undermine trust, highlighting the need for clearer communication and data transparency. This study broadens theoretical understanding by extending ELM and EVT to the context of GAI-driven sustainability efforts, providing an integrated framework that encompasses cognitive and motivational trust drivers. These insights fill gaps in technology adoption research and offer practical guidance for developing GAI platforms that effectively support pro-environmental behavior change.