Tackling tricky complaints: the impact of AI agents and intention hiding strategies on user responses

Purpose – This research serves a twofold purpose: first, to identify and categorize two common intention hiding strategies used by frontline employees when handling tricky user complaints, specifically evasive hiding and rationalized hiding; and second, to systematically examine the interactive effects of agent type (AI vs. human) and these strategies on users’ willingness to forgive.

Design/methodology/approach – Three experiments (N = 820) were conducted to investigate how agent type (AI vs. human) interacts with different intention hiding strategies to influence users’ willingness to forgive. The experiments also tested the mediating effects of perceived negative motives and perceived sincerity, exploring how AI capabilities (mechanical vs. thinking) shape user reactions.

Findings – Users exhibit a higher willingness to forgive AI agents than human agents when an evasive hiding strategy is used; conversely, human agents elicit more favorable responses when employing a rationalized hiding strategy. These effects are mediated by perceived negative motives and perceived sincerity. Furthermore, mechanical AI agents are more effective when using evasive hiding strategies, whereas thinking AI agents perform better with rationalized hiding strategies.

Originality/value – This research extends service recovery theory by introducing evasive and rationalized hiding as intention hiding strategies and by demonstrating that user responses vary according to the alignment between agent type and hiding strategy type. The findings also enrich research on mind perception and AI interaction by uncovering the underlying psychological processes and highlighting the influence of AI capability design on users’ interpretations.