Mechanism of Citizen Evaluation of Policy Using Machine Self-Learning

Abstract

Public policy is designed to influence society. In turn, members of society - citizens should evaluate this policy so that society can control it. According to the concept of digital society, policy evaluation is based on artificial intelligence solutions, primarily machine learning. In addition, when monitoring politics by society, it is necessary to take into account the human factor. The article discusses theoretical and practical issues that arise when a citizen applies the machine learning procedure to determine alternative evaluations - ratings of a politician. In conditions of uncertainty, it is assumed that this politician knows his ability to meet the needs of society better than the citizen. Using this knowledge, the politician can manipulate own activities in order to get higher ratings today and in the future. Such undesirable activity can lead to the failure to use the available opportunities in which the citizen and society as a whole are interested. To solve this problem in conditions of uncertainty, a mechanism for citizen evaluation of politician is proposed. This mechanism includes the procedure of machine learning of dichotomy and the formation of alternative ratings of a politician. Sufficient conditions have been found for the synthesis of such a mechanism in which a politician fully uses the existing opportunities in the interests of the citizen and society as a whole. The functioning of this mechanism is illustrated by the example of an evaluation of the national policy on vaccination against COVID-19 in the UK. Such a mechanism encourages the politician to use all available opportunities in the public interest. The developed mechanism can be used by any citizen for permanent evaluation of policy using machine self-learning. For this, for example, such mechanism can be implemented as an application on a smart phone.