Generative AI and sustainable consumption: assessing cognitive and motivational drivers of trust through the elaboration likelihood model and expectancy–value theory

Purpose Grounded in the elaboration likelihood model (ELM) and expectancy–value theory (EVT), this study aims to investigate how user trust fosters the adoption of generative AI (GenAI) sustainability recommendations. It evaluates the impact of central cues (information quality), peripheral cues (anthropomorphism) and expectancy–value factors (perceived ease of implementation and personal relevance) on trust and examines whether perceived information complexity moderates these relationships. Design/methodology/approach The authors tested a unified ELM–EVT framework using survey data from 673 GenAI users in Indonesia. Data were analyzed using a structural equation modeling approach in SmartPLS 4.1. Findings High-quality, transparent information and anthropomorphic design significantly enhance trust, which in turn increases intentions to adopt GenAI sustainability recommendations. Trust is further strengthened when recommendations are easy to implement and personally relevant, whereas perceived information complexity has no significant moderating effect. Originality By integrating ELM and EVT in a sustainability context, this study offers a model for explaining GenAI recommendation adoption. Specifically, the authors unify central (information quality) and peripheral (anthropomorphism) routes with motivational drivers (ease of implementation, personal relevance) in a single model of trust formation and test perceived information complexity as a boundary condition. The results guide designers toward user-centric systems that balance clarity, anthropomorphic engagement and motivational alignment to advance sustainable consumer behavior.