Political Preferences for AI: Comparative Evidence

• In Event: Voters and Communication in Advanced Democracies

Abstract

Artificial Intelligence has been proposed as a key factor in future technological and social development. It might be a key driver of future inequality. There is a wide range of alleged effects of AI on society, and compounded to these effects, multiple issues are identified by experts such as automation and inequality, ethical judgment, algorithmic discrimination and augmentation. Yet as to how the public perceives these components among each other and prefers which AI policy is an open issues. We test a novel hypothesis that the political system and context affect the importance individuals place on different dimensions of AI technological systems, especially taking into account potential inequality and augmentation. We theorize among dimensions of the institutional clarity of responsibility attribution, competition and political value systems, and show how these components change preferred AI policies via context dependent political preferences. For this purpose we present novel experimental evidence of conjoint studies on AI system policy and regulation across five different countries: the US, the UK, Chile, India and China.

Author

• Soenke Ehret, Nuffield College, University of Oxford