Abstract

Giving AI speakers with opportunities for cognitive and emotional decision-making leads to issues of trust in AI speakers. Computers Are Social Actors paradigm and attribution theory can explain the relationship between trust targets and identify which factors affect trust in them. We focus on the cognitive and emotional decision-making of AI speakers, and aim to find out whether people trust AI’s decision and why they trust it. We conducted a survey, and the results showed that perceived autonomy and perceived empathy of AI speakers have a very positive effect on trust in AI speakers. Also, people trust AI speakers because they are human-like. Finally, people who want conversational AI speakers with social communication trust AI more when they perceive empathy in AI speakers. Results are meaningful in extending the CASA paradigm, which can establish social relationships with computers, to connections with AI speakers.