Differences in acceptance of AI and human judgment in indirect reciprocity

Abstract

With the rapid development of Artificial Intelligence (AI) related technologies, the coexistence of humans and AI has become a significant issue worldwide. The spread of AI is not limited to the automation of services but is also applied to evaluating human behavior. Evaluation of human behavior is an essential issue for human society. Because a mechanism called indirect reciprocity, in which people evaluate each other's actions and share their reputations is a fundamental mechanism for maintaining large-scale human cooperation. Here, we focus on justified defection in indirect reciprocity situations. We analyze how the degree of acceptance of justified defection differs between human and AI evaluations and the factors that determine the difference. The results showed no difference between the degree of acceptance of AI judgments and human judgments, but those who were more concerned about AI rated negative judgments by AI lower and positive judgments by AI higher than those who were less concerned about AI.