2022
By Sonia Jawaid Shaikh
Artificial intelligence (AI)-enabled technologies are increasingly being used by humans to make decisions across a variety of settings. The basic idea underlying this technological setup is to provide humans with some kind of machine-based recommendation (e.g. employability score, risk score) which can be utilized in making a judgment or decision. An important consideration in these circumstances has to do with overconfidence i.e., the extent to which humans systematically overestimate accuracy of their decisions and precision of their knowledge. Overconfidence negatively affects various organizational outcomes. However, it remains unclear how and to what extent it evolves when human decision-making is influenced by machine recommendations. This project investigates antecedents, extent, and correction of overconfidence within the context of machine-driven recommendations for decision-making. A series of experiments on a virtual platform will investigate how interaction with AI affects overconfidence and also explore design-based solutions to reduce it.