How should intelligent agents apologize to restore trust? Interaction effects between anthropomorphism and apology attribution on trust repair

• Investigated the trajectory of a fintech AI agent’s trust repair process. Interaction effect anthropomorphism and apology type on was found. In human-like agent, internal attribution repaired better. machine-like external A computer-like agent with an most harmful at repair. Trust is essential in individuals’ perception, behavior, evaluation intelligent agents. Because, it primary motive for people to accept new technology, crucial when damaged. This study investigated how agents should apologize recover effectiveness different compared based two seemingly competing frameworks Computers-Are-Social-Actors paradigm automation bias. 2 (agent: Human-like vs. Machine-like ) X (apology attribution: Internal External) between-subject design experiment conducted ( N = 193) context stock market. Participants were presented scenario make investment choices artificial intelligence advice. To see initial trust-building, violation, process, we designed game that consists five rounds eight (40 total). The results show more efficiently apologizes rather than attribution. However, opposite pattern observed among participants who had agents; condition showed better Both theoretical practical implications are discussed.