Control and the participation dilemma of artificial intelligence in public administration: An evaluation using conditional probability calculations

Abstract

AI and algorithms are now affecting daily life, communities, and government structures more than ever before. AI is an advanced information and communication technology tool that can transform both participation within a public administration and the structure of public administration. However, it is critically important to consider the relationship between AI control and participation functions, which has not been adequately addressed in the literature. For this reason, this study evaluates AI control and participation functions in terms of the concept of 'govermentality'. AI, which constitutes the main axis of digitalization today, has begun to make its presence felt in the digital transformation of public administration. This study examines whether AI is more likely to be used to control the governed or to participate in management. The study will use Foucault's concept of 'governmentality', and the related concept of 'control', which Chatterjee, who has made significant contributions to the study of governed politics through Foucault's ideas, sees as a fundamental aspect of governmentality, as a theoretical framework. Political power, or administration, is the agent that can make artificial intelligence the object of digitalization in public administration. However, there is no indication that this state of being a subject will be shared (without struggle) with citizens. Undoubtedly, Foucault emphasizes the possibility of resistance within domination and domination within what is considered freedom. In this context, even the use of AI for control purposes may allow participation (and vice versa). Therefore, within the study, control and participation (control within participation, participation within control) have been shown to influence each other to varying degrees. To illustrate the possible outcomes of this interaction, we applied conditional probability calculations. Within the scope of our study, we found a dramatic difference in favor of control in the probability calculation that is made by expressing control, participation, and the interaction of these two factors in percentage units.