Data-driven Democracy? Automated Decision-making, Difference and Preferences

• In Event: New Directions and Emerging Technologies

Abstract

The mass adoption of the Internet has produced a deluge of data, a phenomenon termed as the 'new oil' of the digital economy. When this data is combined with algorithms, it enables the creation of computer systems capable to take decisions depending on context. Initially used as 'expert' systems in controlled environments only, they are now starting to enter the public space in the form of devices such as self-driving cars or drones, but also increasingly as part of decision making systems with direct political effects. For example, public opinion can be influenced through the algorithmic selection of information shown in their news feed to match their preferences. This works similarly as online stores offering products based on prior purchases and search terms, but instead of affecting just purchase decision they can affect voting, therefore opening the possibility of electoral manipulation. Similarly, machine learning techniques are combined to produce 'recidivism scores' which are used by judges to assess their decisions. These technologies are increasingly seen as the source of destabilizing phenomena in contemporary democracies such as political polarisation, hate speech and the emergence of extremist and authoritarian regimes. In this paper, we evaluate the destabilization potential of algorithmic decision making looking at the normative foundations of the theory of democracy, focusing on the contributions of preference formation with an emphasis on diversity and inclusion.

Authors

• Jesus Manuel Benitez

• Margarita Gomez-Reino, Universidad Nacional de Educación a Distancia