Systematic review maps generative AI’s double-edged role in tourism and hospitality

Generative artificial intelligence is reshaping how tourism information is produced, exchanged, and consumed, but a new systematic review warns that its benefits arrive bundled with systemic risks that could undermine the interpersonal authenticity on which hospitality depends. The review, conducted by researchers at Macao Polytechnic University and published in Frontiers in Artificial Intelligence, synthesizes a fast-growing but previously dispersed literature into a single evidence-based framework.

Following the PRISMA 2020 reporting protocol, the research team searched the Web of Science Core Collection and Scopus databases and identified 3,825 records. From this pool, they carried out a thematic synthesis of 461 publications, organizing the evidence around three questions: where generative AI creates value on the supply and demand sides of tourism and hospitality, which ethical risks and organizational dilemmas accompany its adoption, and how machine intervention conflicts with interactional authenticity in service encounters.

The synthesis concludes that generative AI operates as a double-edged sword.

Yet these gains coexist with significant hazards documented across the reviewed literature. The authors highlight algorithmic bias and hallucination, in which models generate skewed or fabricated outputs, alongside privacy trade-offs that arise when personalization depends on extensive data collection. They also identify employee deskilling as an organizational dilemma, suggesting that heavy reliance on automated systems may erode the human capabilities that service work traditionally cultivates.

The review’s third strand addresses the tension between machine intervention and authenticity in service interactions. Machine-generated empathy, the authors note, is computational in essence, and this quality can elicit expectancy violations and algorithm aversion among guests.

To integrate these strands, the review proposes an efficacy-risk hedging framework embedded in a socio-technical account of human-AI collaboration. The framework links application-level benefits, systemic risks, and the authenticity conflict within one structure, with each claim anchored in article-level evidence counts rather than anecdotal observation.

From this framework, the authors derive a tiered human-in-the-loop governance model, positioning human oversight as a graduated mechanism rather than a binary choice between full automation and full human control.

The authors frame the work as a contribution to a literature whose application scenarios, ethical consequences, and authenticity implications had remained scattered across a rapidly expanding body of research. By consolidating 461 publications into a coherent structure, the review offers a map of where generative AI is already creating value in tourism and hospitality, where its risks concentrate, and where future research agendas should focus as machine-generated content and machine-mediated service encounters become more common across the sector.

Subject of Research: Applications, ethical risks, and authenticity implications of generative artificial intelligence in tourism and hospitality

Article Title: Applications, ethics, and future agenda of generative artificial intelligence (GenAI) in tourism and hospitality: a systematic review

Article References: Ho, W. I., Yin, Y., Yu, Y., Hu, H., Cai, J., & Wang, C. (2026). Applications, ethics, and future agenda of generative artificial intelligence (GenAI) in tourism and hospitality: a systematic review. Frontiers in Artificial Intelligence, 9, Article 1935676. https://doi.org/10.3389/frai.2026.1935676

Image Credits: AI Generated

DOI: 10.3389/frai.2026.1935676

Keywords: generative AI, tourism, hospitality, systematic review, algorithm aversion, algorithmic bias, hallucination, privacy, deskilling, human-in-the-loop, service authenticity, PRISMA