Increasingly, decision-makers rely on algorithms to make decisions or delegate tasks. However, depending on their perception of these algorithms, they may adopt irrational behaviors. For instance, they may prefer to delegate certain tasks to humans who are less efficient than algorithms (Meehl, 1954; Dawes, 1979; Önkal et al., 2009; Dietvorst et al., 2015). These studies highlight that decision-makers perceive algorithmic assistants or intermediaries differently from human counterparts. In this presentation, we will examine a series of experiments that demonstrate the impact of algorithm perception on human decision-making.
First, we will explore the idea that decision-makers can be overly credulous when following recommendations from certain algorithms. A series of three experiments will illustrate this phenomenon under different conditions. Next, we will investigate the quality and level of trust placed in an algorithm when it provides recommendations. Finally, a concluding study will focus on the perception of algorithmic responsibility. The objective will be to determine who is perceived as responsible when an algorithm makes an inequitable decision.

Practical information