Responsiveness to a Digital Age: Voting on Artificial Intelligence in the European Parliament

Abstract

Artificial Intelligence is a growing technology that is likely to be central for economies and industries across the whole globe. The European Union wants to be at the forefront of the development and deployment of this technology. To lead the development of artificial intelligence is essential to keep up with superpowers like USA and China. The problem is that artificial intelligence can also harm people if not deployed and developed in a human-centric way that benefits the public. It is therefore essential to understand if the EU is responding to public preferences regarding the development of this technology. This thesis investigates if the Members of the European Parliament are responding to the public opinion on Artificial Intelligence. The theoretical argument originates from the theory of responsiveness, which argues that policymakers move their policy preference along with public opinion. The thesis also tries to uncover what might condition responsiveness on this issue. This thesis uses a multilevel analysis method to test the argument. The analysis combines roll-call voting data from the European Parliament with public opinion survey data making it possible to examine effects on multiple levels. The empirical analysis shows no discernible effects between MEPs and the issue of artificial intelligence. However, it also shows that the largest determining of the votes of the MEPs are their national parties. So, if the MEPs vote in line with their national parties, the responsiveness might be found on the party level. The party-level analysis shows that party size significantly affects the parties' votes, and smaller parties seem more responsive. By including a cross-level interaction between party size and public opinion on AI, the analysis finds a significant effect of the interaction. The interaction effect indicates responsiveness to public opinion on AI conditioned that the party is small enough.