Disentangling fairness, accountability and transparency in AI travel recommendations: the role of algorithm trust across hedonic and utilitarian contexts

Purpose This study aims to examine how tourists’ perceptions of fairness, accountability and transparency (FAT) in artificial intelligence (AI) recommendation algorithms shape algorithm trust and how trust influences recommendation acceptance and resistance, with tourism product type as a moderator. Design/methodology/approach A scenario-based online survey of 355 Korean adult consumers was conducted via travel-related Naver cafés and KakaoTalk communities. Hypotheses were tested with structural equation modeling in AMOS 23.0, and the moderating role of tourism product type was assessed through multigroup analysis. Findings All three FAT dimensions positively predicted algorithm trust, with transparency showing the largest effect, followed by accountability and fairness. Algorithm trust positively predicted recommendation acceptance and reduced resistance. Indirect effects through trust were significant for accountability and transparency but not fairness; accountability influenced trust only in the utilitarian condition. Research limitations/implications Self-reported, cross-sectional data from a Korean-only sample may limit causal inference and generalizability. Future research could use objective behavioral data, longitudinal designs and cross-cultural samples to extend the FAT–trust–behavior framework. Practical implications Tourism platforms should prioritize explainable AI features, adopt context-sensitive communication – emphasizing accountability cues for utilitarian services and transparency-based emotional appeal for hedonic ones – and focus on building trust at the first AI recommendation touchpoint. Social implications The findings provide empirical grounding for tourism-specific AI governance. The dominant role of transparency suggests that policymakers should consider introducing recommendation rationale disclosure requirements for AI-powered tourism platforms, enabling consumers to understand why specific products are suggested. In addition, the finding that consumers cannot directly evaluate algorithmic fairness highlights the need for independent, third-party fairness audits of recommendation systems to protect consumers from potential algorithmic bias that may systematically favor certain destinations, accommodations or price segments over others. Originality/value This study is one of the few to empirically apply the AI ethics FAT framework to tourism consumer behavior. The findings suggest that the three FAT dimensions may not contribute equally to algorithm trust – with transparency showing the strongest association regardless of context and accountability becoming relevant primarily in utilitarian service settings – which points to the limitations of treating FAT as a uniform construct in AI recommendation design. The resulting framework offers tourism platforms an empirical basis for tailoring AI trust strategies to specific product categories.