The adoption of algorithmic decision-making agents over time: algorithm aversion as a temporary effect?

Abstract:

Many individuals encounter algorithmic decision-making agents with algorithm aversion – the irrational discounting of superior algorithmic advice. So far, we know little about how algorithm adoption develops over time and how people may overcome algorithm aversion. In response, we explore the factors that foster the adoption of algorithmic decision-making agents – initially and over time. Based on an experiment with incentive-compatible awards over several rounds, we find that one’s knowledge about peers successfully using the technology as well as low transaction costs serve as strong initial motivators to foster initial algorithm adoption. Further, by revealing that adoption rates increase and initial difference in adoption rates become smaller over time, we find evidence that despite the technology’s particularities, algorithm aversion seems to have a temporary effect only.