The Role of Algorithm and Task Familiarity in Algorithm Aversion: An Empirical Study

Abstract

Algorithm aversion, characterized by the tendency to distrust algorithmic advice despite its demonstrated superior or identical performance, has become an increasingly concerning issue as it reduces the practical utility of algorithms. To gain insights into this phenomenon, our research centers on individual traits, specifically focusing on familiarity with algorithms and familiarity with the task at hand, and their connections with attitudes toward algorithms. We construct a causal model to delve into these relationships and assess how attitudes, in turn, impact algorithm aversion. Our analysis draws upon data collected through an online survey involving 160 participants, and we employ PLS-SEM for our analysis. The results underscore a noteworthy positive correlation between familiarity with algorithms and attitudes toward algorithms. Interestingly, our findings indicate that familiarity with the task or domain knowledge does not significantly influence attitudes. Moreover, attitudes are demonstrated to have a negative impact on algorithm aversion. These discoveries hold significant implications for comprehending and addressing the issue of algorithm aversion. They shed light on the roles of individual traits and attitudes in shaping people's acceptance of algorithms, ultimately offering valuable insights for mitigating this phenomenon.