How Artificial Intelligence Modifies Political Opinion Formation — An Experimental Approach to Understand the Influence of Political Bots in Online Social Networks

Artificial intelligence (AI) modifies how political campaigning in online social networks is implemented. AI-enhanced automation allows scaling up quantity and frequency of content and messages. Algorithmic personalisation by means of micro-targeting and human imitation initiate sophisticated and optimised forms of political campaigns online. In recent years, these AI technologies have been adopted in the form of so-called political bots. Various political actors instrumentalise political bots to propagate information, sometimes also misleading content such as disinformation, in order to influence the political opinions of individuals. However, the actual causal impact of political bots on political opinion formation on the individual as well as at the group level and the respective underlying influence mechanisms are not well understood. This dissertation project therefore examines the technological features inherent to the functioning of a political bot to understand and evaluate the impact of bot-behaviour on political opinion formation by conducting lab and online experiments. Hypotheses derived from political communication theory and opinion dynamics models are tested and computational simulations are implemented in order to theorise AI-modification of political opinion formation in online social networks.

Keywords: Artificial Intelligence, (Social/Political) Bots, Opinion Dynamics, Political Communication, Political Campaigning, Social Media, Causal Inference, Computational Social Science