Trust and Acceptance of Social Robots: Discussion & Conclusion

cover
12 Jun 2024

Author:

(1) Katrin Fischer, Annenberg School for Communication at the University of Southern California, Los Angeles (Email: katrinfi@usc.edu);

(2) Donggyu Kim, Annenberg School for Communication at the University of Southern California, Los Angeles (Email: donggyuk@usc.edu);

(3) Joo-Wha Hong, Marshall School of Business at the University of Southern California, Los Angeles (Email: joowhaho@marshall.usc.edu).

Abstract Introduction & Related Work

Method

Analysis & Results

Discussion & Conclusion

References

IV. DISCUSSION & CONCLUSION

This study (n = 239) was conducted to explore the impact of first impressions and trustworthiness characteristics on social robot trust and acceptance. The results of the mediation analysis show that while there was no direct effect of first impressions on trust, the indirect effects of both warmth and competence on trust via the mediator trustworthiness were significant. Separating trustworthiness into its sub scales ability, benevolence and integrity run as parallel mediation confirmed that both ability and benevolence, but not integrity, produced significant indirect effects on trust. Other research [7] has documented this phenomenon and suggested that of these three sub scales, integrity only predicts trust via ability and benevolence. A serial parallel model placing integrity before ability and benevolence, which in turn predicted trust, confirmed this. Finally, a serial parallel mediation model including first impressions, trustworthiness constructs, trust and behavioral intention per the UTAUT model was devised and analyzed. All indirect effects were significant, indicating that participants whose first impression of the social robot was positive were likely to intent to use if they also perceived it as trustworthy and had trust in it. Notably, in this final model integrity was placed in parallel with ability and benevolence and produced a significant effect predicting intention to use via trust, while a prior model showed that it predicted a trust outcome only in serial with ability and benevolence, indicating that predicting intention to use may rely on all three subscales to a higher degree than trust does. This model gives evidence that stereotypical first impressions, trustworthiness, trust and use behavior are closely related and informed by each other. We recommend further investigations into how real-world interactions may influence the reported results and how contextual changes and different user segments may affect the role of trustworthiness characteristics.

This paper is available on arxiv under CC 4.0 license.