AI Governance

AI governance: How can politics shape artificial intelligence?

There has so far been a lack of research on the possibilities open to politics to intervene in, control and shape the field of artificial intelligence (AI). The project therefore examines and systematizes previous approaches of political governance and is based on existing initiatives and expertise at federal and state level.

Artificial intelligence (AI) has become a political issue in Germany at the latest since the federal government’s current coalition agreement, with the number of expert reports and newly founded committees commissioned by politicians increasing as a result. Some federal states have already developed political initiatives on the issue – for example, as part of the Digital Strategy.NRW, the lead project KI@BW, and the AI initiative of the Bavarian State Ministry for Digital Affairs.

A systematic investigation of opportunities to intervene in, control and shape AI beyond technical solutions, and of how various political measures have been implemented in the new policy field of “artificial intelligence”, has so far failed to materialize. The research project “AI Governance” addresses precisely these issues, drawing on existing political initiatives and expertise to answer the following questions:

• Which concrete political measures are being developed and implemented from the various projects, initiatives and strategies in the field of “artificial intelligence”?
• Which political actors are taking responsibility for the new policy field, and what does coordination between the federal and state levels look like?
• What recommendations for developing and implementing instruments and institutions can be made on the basis of the results?

The problem of coordinating such initiatives in a new policy area is not new – confusing constellations of actors have recently emerged at both federal and state level in the area of general network and digitalization policy, too. Although there are studies in political science that deal explicitly with the emergence of a new policy field (Greef 2017; Reiberg 2018), it is not yet possible from the available studies to identify clear rules and patterns in the formation of concrete structures, committees, and approaches to action.

Visualizing previous developments

The aim of the project is to make visible the development so far of an “AI policy” in its different dimensions, to analyze relationships of cooperation and competition, and to investigate whether the control and regulation of artificial intelligence should be understood as an independent policy field. Two successive methods address the research questions: first, the systematic stocktaking of existing AI initiatives in politics; second, personal interviews with experts.

Taking stock of existing AI initiatives

• Complete survey of all previous committees, strategy papers, expert opinions, and catalogues of measures at state and federal level
• Content analysis of relevant documents in preparation for the expert interviews

Interviews with experts

• Identification of experts who play key roles in the discourse
• Group classification: politicians, commissioners, practitioners, actors in civil society

Relevance for (political) practice

The findings of the study will lead to a policy paper addressed to members of the state parliament and employees of the ministerial bureaucracy, primarily in North Rhine-Westphalia. The paper will contain both the key points of the systematic stocktaking and a condensed reflection and summary of the statements of the interviewed experts. On this basis, recommendations for action will be developed to help the actors involved in the further development and implementation of AI governance in the context of the NRW digital strategy.