The information sidedness effect in travel decisions with ChatGPT: when, why, and for whom two-sided recommendations outperform one-sided recommendations

As reliance on generative AI for travel recommendations grows, understanding the factors shaping acceptance is crucial. Drawing on message sidedness and the CASA paradigm, we argue that two-sided recommendations from ChatGPT – balancing positives and negatives – boost visit intentions and credibility more than one-sided ones. Five studies support this, showing that two-sided messages enhance credibility and are especially effective when ChatGPT is perceived as less human-like, under low time pressure, and when AI rather than humans provides recommendations. However, effects may backfire if negatives are too salient, particularly for fixed-mindset consumers. This research advances persuasion theory and practical AI travel strategies.