Whose journey matters? Investigating identity biases in large language models (LLMs) for travel planning assistance

As large language models (LLMs) become increasingly integrated into the tourism and hospitality industry, concerns persist regarding their fairness in serving diverse identity groups. Grounded in social identity theory and sociotechnical systems theory, this study examines ethnic and gender biases in LLM-generated travel recommendations. Using a mix-method fairness-probing approach that integrates machine learning, classical statistical testing, and qualitative analysis, we analysed outputs from four LLMs, including two open-source and two closed-source LLMs, and compared results generated from both semi-structured and unstructured prompts. Results indicate that while hallucinations were effectively mitigated in the LLMs, ethnic and gender biases persisted across both sourced LLMs. Moreover, the findings were consistent between outputs generated from semi-structured and unstructured prompts. The study underscores the need for effective bias-mitigation strategies to enhance the inclusivity, transparency, and reliability of generative AI-driven travel planning systems.