Engineering a social contract: Rawlsian distributive justice through algorithmic game theory and artificial intelligence

Abstract

AbstractThe potential for artificial intelligence algorithms and game theory concepts to offer prescriptive and decision-making capability for humankind is increasingly recognized. This derives from the increasing availability of granular, multivariable, well-curated data offering analytical insights for necessarily complex human behaviors and activities. Of the multitude of situations that this decision-making aptitude presents, the application to governmental policy offers a commanding case. This would allow decisions to be made for the benefit of societies and citizens based on rigorous objective information devoid of the traditional approach of choosing policies and societal values based on the opinion of a handful of selected representatives who may be exposed to a lack of comprehensive data analysis capacity and subject to personal biases. There would need to be a critical requirement of wider socially responsible data practices here, beyond those of technical considerations and the incorporation of wider societal fairness approaches. Amongst the schools of political thought particularly acquiescent to the application by this approach would be the egalitarian approach of John Rawls. Here an Original Position’s pre-determination tool of Veil of Ignorance and ensuing Difference Principal presents a method of distributive justice that can be clearly mathematically defined in economics theory through Wald’s Maximin principle. This offers an opportunity to apply algorithmic game theory and artificial intelligence computational approaches to implement Rawlsian distributive justice that are presented and discussed. The outputs from the algorithmic acquaintance of Rawlsian egalitarianism with applicable state data, protected with appropriate privacy, security, legal, ethical and social governance could in turn lead to automated direct governmental choices and an objective Social Contract for citizens of digitally literate nations.