How marketing mechanisms build trust in AI recommendation systems: a serial mediation model in live-streaming commerce

This study investigates how marketing mechanisms influence consumer trust in AI recommendation systems and purchase intentions in live-streaming e-commerce. Drawing on Social Cognitive Theory and Theory of Planned Behavior, we examine three marketing mechanisms—celebrity endorsement, electronic word-of-mouth, and social media marketing—and their effects on purchase intentions through serial mediation via trust in AI recommendation systems and consumer attitudes. Using survey data collected from July to September 2025 from 330 consumers aged 18–35 in China’s Greater Bay Area, partial least squares structural equation modeling was employed to test the proposed relationships. The results reveal that celebrity endorsement and electronic word-of-mouth significantly build trust in AI recommendation systems, while social media marketing does not show significant effects. Trust in AI recommendation systems strongly predicts consumer attitudes, which in turn drive purchase intentions. Serial mediation analysis confirms the sequential pathway from marketing mechanisms through trust and attitudes to purchase intentions, with the model explaining 42.8% of variance. This research demonstrates that algorithmic trust develops through observational learning from credible human sources rather than generic marketing communications. The findings suggest that e-commerce platforms should prioritize leveraging endorsements and word-of-mouth communication rather than relying on social media marketing activities.