Whom We Trust More: AI-driven vs. Human-driven Economic Decision-Making

Abstract

AI as a new direction in the study of human-computer interaction requires a new look at trust as a phenomenon. In our study, we focus on examining trust in the context of economic behavior. The study took place in two stages. At the first stage, during the interview, we have identified the main factors of trust and mistrust in AI and the specific factors of trust in AI in economic decisions. Also, we have revealed a subjective indicator of the level of trust in the advisor’s recommendations - the economic activity of the participant when performing the recommended action. At the second stage, an experiment was carried out. The participants were asked to play a stock exchange game. The goal of the game was to make money by buying and selling shares. There were an option to ask an advise. For the experimental group, AI acted as an advisor, for the control group, a person (an expert in trading). According to the analysis of 800 economic decisions, economic activity during the game was higher among the participants in the control group who followed the advice of the person (t = 3.646, p <0.001). As a result of the study, three main conclusions were obtained: 1) the level of trust in councils in an economic decision can be expressed in the form of economic activity; 2) the level of trust in economic recommendation depends on whether the recommendation is made by a human or an AI; 3) the specific factors of trust in economic decisions are highlighted: the individuality of the council and the speed of the requested solution.