Effects of prior experience, gender, and age on trust in a banking chatbot with(out) breakdown and repair

Abstract

Trust is an attitudinal construct that can be sensitive to prior experience, gender, and age. In our study, we explored how trust in a banking chatbot might be shaped by these user characteristics. Statistical analysis of 251 participants, who interacted with one of six chatbots defined by humanlikeness (high/low) and conversational performance (no breakdown, breakdown with repaired, breakdown without repair), showed that the user characteristics of gender and age did not significantly impact trust, but prior experience did. Trust resilience was found across the gender and age groups. The effect of users’ prior experience on their trust in a chatbot which they have never used holds implications for research and practice. Future studies on the effect of cultural context, longer interaction episodes, and more diverse application contexts on trust in chatbots are recommended.