Tracing the legitimacy of Artificial Intelligence: A longitudinal analysis of media discourse

Artificial Intelligence (AI) is one of the most relevant technologies of our time. During the last decade, AI has made major technological breakthroughs most recently in the space of generative AI. This development is enabled by an increase in computing power, a decrease in its price, and the emergence of ubiquitous computing, resulting in vast amounts of storable and processable data. However, the diffusion of AI depends on its legitimacy in society, whereby legitimacy is understood as the congruence between organizational activities and their cultural environment. This study aims at understanding the process whereby AI is being (de)legitimated across key industries and over time. To capture and trace the process of legitimation, we rely on media coverage as a form of societal discourse reflecting the legitimation process of AI. We find that the legitimation process gathers momentum in the mid-2010s and the legitimacy of AI increases over time. Furthermore, we identify four types of legitimacy discourse, which integrate sentiment, specific media frames, and appeals of persuasion. Uncovering the four types of legitimacy discourse, we aim at supporting organizations seeking to understand the legitimacy of specific AI applications and how these legitimacy judgments can shift.