Policy Making Based on Real-Scale Social Simulations

In this paper, we discuss decision-making processes for our communities using social simulation tools using some machine learning or artificial intelligence techniques. We take an example of considering preventive measures based on simulation results during COVID-19 pandemic. To avoid explosive infection in Japan, several preventive measures were considered. Among them, short-time business for restaurants, tourism support policies and vaccination schedules are included. I am involved in Covid-19 AI & Simulation Project Team (AISP) of Cabinet Secretariat, Japanese government. My contribution is to provide synthetic population data for real-scale social simulations for specific areas. In those simulations, we do not aim to predict a precise number of infected or severe patients by COVID-19 but to show several simulation results under various scenarios with different simulation parameters. After their simulation results are compared with each other, common outcomes are extracted from their results, and finally they are provided to the government. In that decision making process, their simulation results under several scenarios are shown to government officers, and final decision makings are left for politicians to decide. Since experts who are not elected are not able to take political responsibilities for their decision making, the AISP team shows several scenarios using their simulation models to support government officers and politicians to make their decisions.