AI and human autonomy: a literature review

The interaction with AI technologies has opened a debate about its impact on a core value and concept, namely human autonomy. However, this debate is largely heterogeneous and unstructured. This is the first review dedicated to AI ethics’ scholarly literature concerning the impact on human autonomy. The aim is to map and structure this debate by surveying its key concepts, topics, and gaps. The first part clarifies how human autonomy is understood in the reviewed literature by identifying its multiple conceptions. This shows how different accounts of autonomy underpin different analyses of the impact of AI technologies. The second part maps and explains the ways AI technologies affect the concept and exercise of human autonomy. Two main sets of concerns emerge: first, AI mediation of human activities influences autonomous deliberation, decision-making, and behaviour; second, AI automation of human activities affects the conditions for autonomous choice and action. This categorisation shows how the impact of AI technologies depends on the system, the context, and the conception of autonomy. This literature review develops a common framework providing a comprehensive understanding of the main concepts and issues at stake, as well as indicating future research directions.