The Impact of Asymmetric AI Assistance on Decision-Making in Social Dilemmas: A Study on Human Augmentation in Economic Games

Abstract

Unequal access to AI augmentation may disrupt human cooperation. In this study, we address the impact of asymmetric AI assistance on decision-making in cooperative economic games, focusing on how AI augmentation influences trust, cooperation, and perceptions of fairness in scenarios involving augmented and non-augmented players. Using the Trust Game and the Prisoner's Dilemma, we conducted experiments in which participants interacted under varying conditions of AI access. We found that while AI augmentation did not significantly alter overall cooperation rates, it shaped social perceptions: non-augmented players viewed augmented counterparts as more competitive and less warm, predicting less cooperation in the Trust Game. These disparities in perception highlight the potential of AI augmentation to subtly influence human cooperation. On a larger scale, the findings emphasize the importance of designing equitable AI system access to prevent social divides and promote cooperation in AI-augmented societies.