Deutsch Intern

    Subproject 1: Conditions for legitimate use of machine learning tools in democratic administrative proceedings

    Many constitutions worldwide share the requirement of Article 20 German Basic Law: “all state authority emanates from the people.” The relationship between the state, its institutions and its citizens is defined by a control mechanism that upholds democratic principles and the legitimacy of state decisions. This extends to public administration and civil servants.  In representative democracy, legitimisation is carried on over different levels, starting from elected representatives of parliament to individual civil servants. In addition, decisions taken by the public administration would need to be within the framework of laws, ordinances and regulations. Public administration is entrusted with the authority to take corrective measures in individual cases on the basis of factual and substantive legitimisation.

    When machine learning tools are put in practice, the question then arises as to what level these could be classified within different levels of legitimisation in the context of the representative democracy. Furthermore, as referred to above in terms of elected representation, the question would be whether approval by the public could bring about legitimacy in the use of machine learning tools allowing for fulfilment of state actions. The principle of representative democracy requires a constant control mechanism that is reflective of the opinions of citizens. Here, an analysis will be made of the state of opinion within representative bodies about the use of machine learning tools. In particular the question of whether the state could, and to what extent should, delegate its decision making to automation.