Multilingual AI chatbots in managing cultural expectations across diverse tourist populations: a mixed-methods analysis of smart tourism cities

Multilingual AI chatbots are increasingly used in global tourism cities, yet existing research has focused mainly on translation accuracy and technical performance rather than their ability to navigate cultural expectations, communication norms, and service expectations among diverse visitors. This study addresses this gap by examining how AI-powered chatbots mediate language barriers, cultural misunderstandings, and contextual appropriateness within smart tourism environments. Focusing on Singapore, Barcelona, and Dubai, three cities with distinct levels of cultural heterogeneity, digital infrastructure maturity, and service philosophies, the study investigates how chatbots respond to culturally specific tourist needs, preferences, and communication styles. A mixed-methods design combines computational analysis of 37,842 real tourist-chatbot interactions with ethnographic observations and semi-structured interviews involving visitors from six cultural regions: East Asia, Southeast Asia, the Middle East, Western Europe, Eastern Europe, and North America. Natural language processing identifies large-scale linguistic and interactional patterns, while qualitative data capture nuanced cultural dynamics that automated analysis may overlook. The study further examines whether chatbots reinforce cultural homogenisation by defaulting to Western communication norms or demonstrate genuine adaptability to diverse cultural frameworks, including directness, formality, sentiment expression, and service expectations. Path analysis shows that interaction quality fully mediates the relationship between cultural factors and tourist satisfaction. Cultural Communication Style has the strongest effect on interaction quality, while Bot Cultural Adaptation improves interaction quality but remains inconsistently aligned with users’ communication patterns.