AI HALLUCINATION: HOW TRUSTWORTHY IS GEN-AI FOR TRAVEL PLANNING? REVIEW OF THE GASTRONOMIC EXPERIENCE IN THE TOURISM DESTINATION OF KIRKLARELI, TÜRKİYE

The study’s objective is to examine the AI hallucination by considering the trusthworthy of the use of Gen AI outputs in travel planning from a tourism perspective.  In the study, gastronomy tourism-based outputs for Kırklareli destination in Türkiye were analyzed using ChatGBT as a Gen AI platform. The findings reveal that Gen AI generates two types of hallucinations: intrinsic and extrinsic. These hallucinations are caused by factors such as incomplete, outdated or inaccurate training data, dynamic structure of the tourism sector, production of outputs based on probabilities, training on large amounts of unsupervised data, uncertain prompt and limited context, training data bias, inability to understand real-world complexities, and model complexity. In the study, hallucinations generated by ChatGPT are categorized into six themes: place and location, menu, geographical mismatch, inconsistencies with cultural and daily life reality, incoherency and repetitions, and terminology hallucination. The findings indicates that, in its current state, Gen AI should not be regarded as a totally trustworthy resource for travel planning, but rather as a supportive resource for users.