One Bad Apple: The Spillover Effect of Algorithmic Aversion

In an era increasingly reliant on algorithms, understanding user interactions with these technologies is crucial. This study explores the phenomenon of algorithmic aversion, particularly the spillover effect—how observing poor performance of one algorithm can impact aversion toward other, unrelated algorithms. We investigate the role of divergent thinking as a potential moderating factor and introduce the novel concept of human-AI collective efficacy as a mechanism via which the spillover effect occurs. This study employs a survey-based experiment, involving various algorithmically augmented tasks to assess the spillover effect of algorithmic aversion. Our research aims to expand the discourse on algorithmic aversion, providing insights into salient human-AI relationships, with implications for both theoretical frameworks and practical applications in developing user-centric algorithms.