The Concentrated City: Effects of AI-Generated Travel Advice on the Spatial Distribution of Tourists

Abstract

The analysis of the spatial location of tourists is essential for effective tourism management. This study explores the potential effects of large language models (LLMs) on urban travel planning. Despite growing academic interest in LLMs, empirical research on their specific impact on urban tourist locations remains limited, even though these models may significantly affect tourist behavior and spatial dynamics. This article compares the location of heritage sites in the city of Barcelona that are traditionally visited by tourists (as identified through Instagram) with those recommended by ChatGPT. The results show that ChatGPT tends to recommend a much smaller and more spatially concentrated number of tourist attractions than those shared on Instagram. The findings indicate that ChatGPT reinforces mainstream representations of cities by prioritizing well-known landmarks, potentially overlooking emerging or local attractions. This simplification can lead to tourist overcrowding and the marginalization of less-visited areas. Likewise, it may entail new needs for the management of urban spaces. Urban planners and tourism managers may need to intervene to redistribute tourist flows in a context where various models of tourist behavior will coexist.

This research was funded by the Departament de Recerca i Universitats de la Generalitat de Catalunya, grant number 2021 SGR 01369 and the Agencia Española de Investigación, grant number PID2021-123063NB-I00.