ChatGPT, where should I go? A qualitative exploration of how large language models are experienced as support for travel planning

Large language models (LLMs) are increasingly used for travel planning. Yet, little is known about how travellers experience and interact with such language models. This qualitative study explores how users employ LLMs to plan trips, drawing on the hedonic/pragmatic model of user experience to examine functional and affective dimensions. We collected data from 104 participants with prior experience using LLMs for travel advice through open-ended questionnaire responses. Thematic analysis revealed three key insights: (1) users value the pragmatic benefits of LLMs, such as efficiency, clarity, and confidence in decision-making, while also appreciating hedonic qualities, including inspiration, enjoyment, and authenticity; (2) prompting strategies vary from highly specific and detailed to exploratory and conversational, reflecting evolving mental models of how LLMs should be guided; and (3) while individual use dominates, collaborative use shows that LLMs can also serve as mediators in group planning, supporting shared decision-making. The findings suggest that LLMs function not only as task-oriented tools but also as co-planners that shape the social and emotional dimensions of travel planning. These insights contribute to an understanding of user-centered adoption of LLMs in tourism and highlight design opportunities for more reliable, personalised, and socially aware Artificial Intelligence (AI) travel assistants.