Abstract

The abstract goes here and should be limited to a single paragraph. The abstract should be informative for a broad audience and provide a “big picture” overview of the article. Most abstracts are from 150 to no more than 200 Artificial intelligence systems are increasingly used in domains such as healthcare, finance, and public administration. A consistent finding across these contexts is that individuals often prefer human to algorithmic advice even when AI systems perform better. Existing explanations typically attribute this preference to conscious evaluations of the fairness, integrity, and expertise of AI systems. While informative, these accounts do not consider the role of unconscious processes in shaping advice preferences. This paper advances a complementary perspective by examining algorithmic aversion through the lens of Terror Management Theory (TMT). We argue that preferences for human versus algorithmic advice are shaped not only by deliberative reasoning but also by unconscious existential concerns. As AI systems increasingly perform tasks associated with judgment, morality, and social interaction, they may threaten symbolic resources that sustain meaning, self-worth, and human distinctiveness. These threats can activate defensive responses that manifest as algorithmic aversion. We identify four sociotechnical characteristics of contemporary AI systems that are likely to evoke existential concern: agentive autonomy, relational embodiment, moral stakes, and operational opacity. We integrate these characteristics into a TMT-based research agenda that outlines directions for future research on algorithmic aversion. By reframing algorithmic aversion as an existentially grounded phenomenon, this paper provides a conceptual foundation for future research on human-AI interactions and highlights the importance of attending to symbolic and unconscious processes that shape responses to AI. words, and they can certainly be shorter.

DOI

10.17705/1jais.01030

Download

Share

COinS