This research project aims to analyze the sociotechnical imaginaries and tools of “algorithmic decision-making”, a notion explicitly anchored in objectivity, automation, optimization and calculation, in the context of public action. This project will proceed in two stages. We first intend to study the different sources and forms of “algorithmic decision-making” as specific control technologies. We will then focus on current debates and recent developments in “algorithmic decision-making” related to the technological and industrial development of artificial intelligence, as well as its various applications in the public domain in France. To do this, we will study the ecosystem of the digital state and producers of algorithmic decision-making devices, as well as a series of specific examples of their use by administrations and public services. We will be particularly interested in the normative visions and the structuring of new “sciences of government” (AI, machine learning, behavioral sciences) centered on the engineering of behaviors in their relationship to the sociotechnical imaginaries and the practical devices of “instrumentarian” power that are currently emerging.