AI, Unethical Decision Making and Moral Disengagement: The Devil is in the Detail

Abstract

The age of artificial intelligence and data has brought and continues to promise undeniable benefits to individuals and society. At the same time, there are concerns about the potential harms which the technology may give rise to if misapplied, and the debate continues as to what type of regulations should be put in place to govern the use of AI and data. This paper proposes that increasing specificity in rules reduces unethical decision making when it comes to the use of AI and data, and that this relationship is mediated by moral disengagement (using situational and dispositional measures). The hypothesis was tested using a mixed-method study using quantitative and qualitative data. The study finds that a combination of general rules and specific rules and examples leads to lower unethical decision making, and that this relationship is mediated by situational moral disengagement. Further, education level and gender were shown to have an influence on moral disengagement – though in slightly different ways. Interesting results from the main study also spurred exploratory analysis which suggests that low moral awareness relating to the use of AI and social media data predicts unethical decision making.

Keywords: moral disengagement, unethical decision making, artificial intelligence, AI ethics, ethics, moral awareness, privacy, data protection

JEL Classification: K00, K10, K20, Y40