抄録
In recent years, generative AI has been widely adopted across various fields. In particular, large language models (LLMs) possess highly advanced text-generation capabilities, and their applications have expanded rapidly. Within the tourism domain, services such as automated travel planning, AI-generated tourism guidance, and conversational tourist information systems are becoming increasingly common, with local governments and tourism organizations actively promoting their implementation. Tourism guidance generated by AI has the advantage of providing information tailored to user input. However, generative AI faces well-known issues such as ",Bias", and ",Hallucination.", Inaccurate descriptions of historical, cultural, geographical,
or facility-related information may mislead users, damage the reputation of local destinations, or even lead to practical problems and disputes. Despite these concerns, only a limited number of studies have examined the characteristics of AI-generated tourism guide texts. Furthermore, it remains unclear how differences among LLMs or assigned personas influence the content of the tourism guidance generated, and how such differences might affect the quality and reliability of tourist information services. To address these issues, the present study generates tourism guide texts using multiple generative AI models and conducts a comparative analysis through text-mining techniques. The objective of this study is to clarify the challenges associated with using generative AI for providing tourism information and to derive insights that may contribute to the development of more reliable AI-based tourism services.
