Overcoming Algorithm Aversion: The Power of Task-Procedure-Fit

Decision-making by Artificial Intelligence (AI)-enabled agents challenges some of the fundamental believes of technology acceptance, both from a decision-maker and a decision-affected perspective. Despite the immense technological potential and rapid developments, finding effective roles for the technological agents and forms of human-technology collaborations in decision-making remains challenging. Prior research shows in parts irrational rejection of superior algorithms (algorithm aversion) and only rare situations in which people prefer algorithmic over human decision-makers (algorithm appreciation). We show that severe negative consequences of decisions dominate people’s perceptions of the decision-maker, whereas technological anxiety and the decision-maker’s nature matter for the acceptance of positive decision outcomes (study 1). Expert decision-makers, while emphasizing a strong negative attitude towards AI-enabled decision-makers, do acknowledge the opportunities and potentials of the technology (study 2). When AI-enabled decision-makers become part of a mixed human-technology decision group, the perceived task-procedure fit fully mediates the negative effect of AI involvement on peoples’ willingness to accept the decision. Algorithm aversion thereby negatively moderates the relationship between AI involvement and task-procedure fit (study 3). The results provide insight into the mixed, but overly negative perceptions of AI-enabled decision agents and reveal how identifying an appropriate task-procedure fit can help to overcome algorithm aversion.