Fair causal effect using Social Welfare Function

Abstract

When dealing with fairness in artificial intelligence, it is crucial to consider the equality of the outcomes of its actions or interventions for humans. In this study, we addressed the problem of equalizing causal effects in specific individual attributes. We developed a method for optimizing this problem. Specifically, we established a model for estimating the effects of interventions using data and applied it to determine the optimal intervention that maximized the general social welfare function, including various equality concepts under constraints. The effectiveness of the method was verified by comparing the experimental results with the data.