What is There to Fear? Understanding Multi-Dimensional Fear of AI from a Technological Affordance Perspective

AbstractFear of artificial intelligence (AI) has become a predominant term in users' perceptions of emerging AI technologies. Yet we have limited knowledge about how end users perceive different types of fear of AI (e.g., fear of artificial consciousness, fear of job replacement) and what affordances of AI technologies may induce such fears. We conducted a survey (N = 717) and found that while synchronicity generally helps reduce all types of fear of AI, perceived AI control increases all types of AI fear. We also found that perceived bandwidth was positively associated with fear of artificial consciousness, but negatively associated with fear of learning about AI, among other findings. Our study provides theoretical implications by adopting a multi-dimensional fear of AI framework and analyzing the unique effects of perceived affordances of AI applications on each type of fear. We also provide practical suggestions on how fear of AI might be reduced via user experience design.Keywords: AIfeartechnological affordanceshuman-AI interactionuser experience AcknowledgementsWe thank the College of Communication Arts and Sciences at Michigan State University for the Brandt Fellowship awarded to Wei Peng for partially supporting the data collection in this work. The authors confirm their contribution to the paper as follows according to the CRediT author statement: EZ: Conceptualization, Methodology, Resources, Investigation, Data Curation, Formal Analysis, Writing-Original Draft, Writing-Review & Editing; MM: Conceptualization, Methodology, Formal Analysis, Writing-Review & Editing, Supervision; MR: Resources, Writing-Review & Editing; WP: Writing-Review & Editing, Funding Acquisition.Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 Li and Huang use the term "AI anxiety" but given that their conceptual model is grounded in fear theories, we consistently use "fear of AI" in this paper.2 Due to the extremely small sample size of non-binary participants, we had to remove the three participants from the final sample. This yields to a sample size of 717.3 The original fear of bias behavior scale has three items, with Cronbach's Alpha at .48. We believe this is because one item does not hold conceptual consistency with the other two, so we only kept the two items.4 We removed fear of against ethics dimension from further data analysis due to the subscale failing to demonstrate reliable internal consistency.5 We conducted an independent sample t-test to test if the CloudResearch sample and student sample are statistically different. Results revealed that there were no significant differences between the two samples in any of the dependent variables of interest.6 Although the VIF statistics indicate we don't have multicollinearity concerns, we also provide a correlation table for all independent variables in Appendix A.Additional informationNotes on contributorsEmily S. ZhanEmily S. Zhan is a PhD student from the College of Communication Arts and Sciences at Michigan State University. Her research focuses on how technology mediates people's needs, motivations, and behaviors in terms of facilitating collective online phenomena. She is also passionate about gender studies.María D. MolinaMaría D. Molina (Penn State University) is an Assistant Professor in the Department of Advertising & Public Relations at Michigan State University. Maria's research explores the social and psychological implications of sharing online, focusing on how we respond to Artificial Intelligence tools that curate user-generated content.Minjin RheuMinjin Rheu (Michigan State University) is an Assistant Professor in School of Communication at Loyola University Chicago. MJ studies and teaches the psychology of how people are influenced by media content, specifically their understanding of self, attitudes, and behavioral decisions.Wei PengWei Peng (University of Southern California, 2006) is a Professor in the Department of Media and Information at Michigan State University. Her research focuses on the psychological and social mechanisms of behavior change and their application in the design of interactive media for health and wellness promotion.