Patrick Dunlop

While algorithmic decision-making technology can greatly improve efficiency at work, it has the potential to promote unethical decision making in humans who work with the technology. In this paper, we test how the moral disengagement mechanism, displacement of responsibility arises, as an outcome of working with autonomy-restrictive algorithmic decision-making (ADM) technology. We draw upon the tenets of the job characteristics model and moral disengagement theory to test how job autonomy-restrictive ADM affects moral disengagement through displacing an individual’s responsibility for unethical behavior. We experimentally manipulated the level of job autonomy afforded by an algorithmic ADM system across two realistic work simulations. In Study 1 (N = 276), we first showed that participants working with autonomy-restricting ADM engaged in higher rates of unethical behavior that disadvantages customers to the benefit of the organization, compared to those who work with autonomy-affording ADM. In Study 2 (N = 256), we used a novel measure of the cognitions associated with displacement of responsibility to show that working with an autonomy-restricting ADM system leads individuals to experience subjectively lower levels of personal responsibility for the negative consequences of their decisions and leads them to displace this responsibility onto a figure of authority, which can be a human or the technology itself. Altogether, our findings provide new insights into how the cognitive process of moral disengagement unfolds, and indicate that job autonomy, as afforded or restricted in an algorithmic decision-making context, is an important consideration when implementing ADM technology.