Fair or Unbiased Algorithmic Decision-Making? A Review of the Literature on Digital Economics

Artificial intelligence (AI) technologies are being used increasingly to automate tasks and decision-making processes, and to predict user behavior. Although AI has been implemented and studied in depth in the computer science, economics and management fields research on AI is relatively new. As digitization lowers the costs of hosting and collecting data, AI and algorithms are becoming more frequent in several sectors and particularly digital environments. While AI has been designed to improve and accelerate information processing, there are serious concerns that algorithmic decision-making could result in unexpected correlations and unintentional biases. This calls for a better understanding of how algorithms can be used and the potential positive and negative outcomes identified in the literature. We review the empirical and theoretical literature highlighting the most critical issues inherent in algorithmic decision-making in the digital economy. We identify the expected and unexpected effects of the use of algorithms, and their application in different sectors. We also discuss the trade-off between fairness and unbiased algorithmic decision-making and provide some practical implications and directions for future research