关键词: 人工智能面试官, 信任, 形象专业性, 求职者反应, 相似吸引理论

Abstract: AI interviews are becoming an increasingly integral component of organizational recruitment processes, leading to more applicants interacting with AI interviewers rather than human ones. In this context, it is crucial to understand how the appearance cues of AI interviewers influence applicants’ reactions. Drawing on similarity-attraction theory, this study proposed that gender congruence between AI interviewers and applicants would enhance applicants’ trust in the interviewer, which in turn reduced their emotional creepiness and privacy concerns. Additionally, the professional appearance of the AI interviewer influenced this process. When the interviewer appeared unprofessional, applicants leaned more on gender-related cues. As a result, the indirect effects of gender congruence on emotional creepiness and privacy concerns—through the aspect of trust—became more significant. To test our hypotheses, we conducted two scenario-based experiments and one experience sampling study. Study 1 employed a 2 (gender congruent vs. gender incongruent) × 2 (high vs. low professional appearance) between-subjects design to examine the mediating role of trust in the relationship between gender congruence and applicants’ emotional creepiness and privacy concerns, and the moderating role of the interviewer’s professional appearance. We recruited 440 participants with interview experience from Credamo and selected the Chinese AI interviewer appearances for the formal study based on a pretest. Participants watched a video simulation of an interview. Study 2 used an experience sampling method involving 112 job-seeking university students from five universities who participated in a 30-day daily survey, reporting on their experiences and perceptions of AI interviews for one month. Study 3 replicated the design of Study 2—a 2 × 2 between-subjects experiment—and examined whether the findings extended to AI interviewers of other racial backgrounds. A separate pretest was conducted to select the non-Chinese AI interviewer appearances used in the study. Study 1 demonstrated that applicants’ trust in the AI interviewer mediated the effect of gender congruence on emotional creepiness and privacy concerns, and that this indirect effect was moderated by the interviewer’s professional appearance. Specifically, when the AI interviewer’s professional appearance was low, applicants relied more heavily on gender-congruent cues, thereby strengthening the positive effect of gender congruence on trust and further reducing emotional creepiness and privacy concerns. When professional appearance was high, this reliance was attenuated. Study 2 further confirmed that gender congruence between AI interviewers and applicants enhances trust in the interviewer, which in turn reduces applicants’ emotional creepiness, while again supporting the moderating role of the interviewer’s professional appearance. However, the within-person variance of privacy concerns did not reach significance in this study, indicating that applicants’ privacy concerns remained relatively stable across different AI interview contexts. Study 3 extended these findings to a cross‐cultural context, showing that even when applicants interacted with AI interviewers of a different racial background, gender congruence similarly fostered trust and subsequently reduced emotional creepiness and privacy concerns. Across studies, the interviewer’s professional appearance consistently moderated these effects, highlighting its importance in shaping applicant reactions to AI interviewers. This research advances the understanding of applicant reactions in AI interviewing by shifting the focus from system functionality to the design of the AI interviewer’s appearance. We identify two key appearance cues—gender congruence and professional appearance—and demonstrate that they influence trust in AI interviewers and subsequent applicant reactions. Second, this research extends similarity-attraction theory to AI recruitment, demonstrating that applicants apply human-like social categorization to AI interviewers: gender similarity enhances perceived similarity and increases trust. Finally, this research enriches the literature on AI anthropomorphism by revealing a compensatory mechanism among appearance cues. A professional appearance acts as a strong cue that anchors trust judgments, diminishing the effect of weaker cues, such as gender congruence. When professionalism is low, applicants rely more heavily on gender similarity to reduce uncertainty.

Key words: AI interviewer, Trust, Professional appearance, Applicant reaction, Similarity-attraction theory