Abstract This opinion paper critically examines the potential and limitations of language model-based AI, with a focus on ChatGPT, in the context of tourism research. While recent studies emphasize the use of generative AI in education and service applications, little attention has been paid to its validity, reliability, and objectivity for scientific knowledge creation in tourism. Through empirical tests involving text analysis of tourism abstracts and image interpretation, the study reveals significant shortcomings in consistency, reliability, and methodological rigor. The findings underscore the risks of overreliance on AI-generated outputs without critical verification and highlight the necessity of maintaining established scientific standards. The paper advocates a cautious integration of AI as a supportive tool rather than a substitute for rigorous research.
Thomas Bausch (Tue,) studied this question.