Defensive Attribution of AI Fairness: The Role of Procedural Fairness in AI’s Moral Responsibility

Although artificial intelligence (AI) can improve human welfare and well-being in many ways, AI has sparked controversies regarding efficiency, bias, and fairness. Motivated by defensive attribution, this study argues that perception of AI fairness and responsibility attribution of biased decisions can be affected by whether they benefit from the decisions or are discriminated against by the AI decisions. This study also examined how individuals’ abstract, or concrete thinking (construal levels) modulate their fairness judgments, an area that remains underexplored despite its importance. Two interconnected experiments employing an experimental-causal-chain approach were conducted with English-speaking adults (18 years or older) residing in the United States, recruited via Prolific, to investigate the effects of construal level and perceived benefits on perceived algorithmic fairness and organizational moral responsibility. Key findings reveal that perceived benefits from AI decisions critically influence fairness perceptions, and that procedural fairness plays a vital role in attributing moral responsibility to organizations. Females showed greater sensitivity to fairness than males when not benefiting from AI decisions. This research enhances our understanding of the psychological factors influencing algorithmic decision-making, providing crucial insights for creating fairer AI decision-making systems and highlighting the importance of procedural fairness in promoting responsible practices within organizations.