This study examines how AI-generated and human-delivered risk warnings influence consumers' purchase intentions toward ultra-low-priced tourism products. Drawing on signaling theory, it investigates the sequential mediating roles of source credibility and perceived risk.
Two experimental studies were conducted in offline and online tourism booking contexts. Study 1 employed a scenario-based experiment involving face-to-face tourism consultation settings, whereas Study 2 used a simulated online travel agency (OTA) booking interface. Partial least squares structural equation modeling (PLS-SEM) and bootstrapping procedures were used to test the proposed serial mediation model.
Across both studies, human-delivered warnings generated significantly higher perceived source credibility than AI-generated warnings. Higher source credibility subsequently increased perceived risk and reduced consumers' purchase intentions toward ultra-low-priced travel packages.
This study advances AI-assisted tourism communication by showing that identical risk warnings acquire different credibility when attributed to AI versus human sources. It further contributes to signaling theory by identifying source attribution as a condition under which otherwise similar warning signals generate different risk evaluations and behavioral consequences.
