AI or humans: Who decides fairly when the stakes are high?

Fairness heuristic theory provides a lens through which to understand how individuals form fairness perceptions in organizational settings. This theory suggests that people use fairness as a cognitive shortcut to assess the trustworthiness and legitimacy of decision-makers.

Artificial Intelligence (AI) has become a critical component of organizational decision-making, influencing key processes such as hiring, performance evaluations, and even disciplinary actions. However, how people perceive the fairness of decisions made by AI compared to those made by humans remains a subject of growing debate. The research paper "For Me or Against Me? Reactions to AI (vs. Human) Decisions That Are Favorable or Unfavorable to the Self and the Role of Fairness Perception," authored by Jungmin Choi and Melody M. Chao and published in Personality and Social Psychology Bulletin (2024), explores this vital topic.

The study examines how decision outcomes - whether favorable or unfavorable - interact with the characteristics of decision agents (AI vs. human) to shape perceptions of fairness. Using fairness heuristic theory as a framework, the researchers conducted six experiments involving nearly 2,800 participants across diverse scenarios. Their findings reveal critical insights into the psychological mechanisms that drive reactions to AI in decision-making contexts.

Fairness heuristic theory: Understanding decision reactions

Fairness heuristic theory provides a lens through which to understand how individuals form fairness perceptions in organizational settings. This theory suggests that people use fairness as a cognitive shortcut to assess the trustworthiness and legitimacy of decision-makers. Choi and Chao's research extends this concept to compare reactions to decisions made by AI versus humans, focusing on how these perceptions shift based on whether the outcome is favorable or unfavorable.

The study hypothesizes that favorable decisions generally elicit higher fairness perceptions regardless of the decision agent, while unfavorable decisions lead to a closer scrutiny of the agent's attributes. In such cases, AI is often viewed as fairer than humans due to its perceived impartiality and unemotional nature.

Key findings and insights

Favorable Decisions and Uniform Fairness Perceptions

Participants in the study consistently rated favorable decisions as fair, regardless of whether AI or humans made them. Positive outcomes, such as receiving a promotion or a reward, overshadowed concerns about the decision-making process. This suggests that favorable results alone can reinforce perceptions of fairness, making the agent's identity less critical.

Unfavorable Decisions: The Critical Role of Decision Agent Attributes

When outcomes were unfavorable - such as being denied a bonus or facing disciplinary action - participants scrutinized the decision-making process more closely. AI emerged as a preferred decision-maker in these scenarios, often being perceived as fairer than humans. This perception was driven by two primary factors:

• Unemotionality: AI was seen as devoid of emotional influence, making it appear objective and free from personal biases often associated with human decision-makers.
• Objectivity: Participants believed that AI relied on data-driven and consistent processes, which enhanced its perceived impartiality.

Fairness Perceptions Influence Behavior

The study revealed that fairness perceptions significantly impacted participants' willingness to accept decisions and remain engaged with their organizations. When AI was viewed as fairer, individuals were more likely to accept unfavorable outcomes and continue investing in their roles. This highlights the potential of AI to mitigate negative reactions when adverse decisions are necessary.

Bias Awareness Reduces Trust in AI

The researchers also explored the impact of raising awareness about AI's potential biases. When participants were informed about the possibility of AI replicating human biases - stemming from flawed training data - their perceptions of AI's fairness decreased. This finding underscores the importance of transparency and education in managing expectations around AI-driven decisions.

Behavioral and organizational implications

The findings from Choi and Chao's research highlight significant behavioral and organizational implications, emphasizing the importance of building trust in AI systems. Organizations can leverage AI's perceived objectivity and impartiality to minimize negative reactions to unfavorable decisions, though this trust hinges on the transparency and reliability of AI systems.

Addressing biases in AI is also critical, as the neutrality of AI is contingent on the quality of its training data. Regular audits and improvements to datasets are necessary to enhance fairness and ensure equitable outcomes. The study also underscores a psychological distinction between how individuals perceive AI and human decision-makers; while humans are scrutinized for biases and emotional influences, AI benefits from an assumption of impartiality.

Furthermore, the research demonstrates that fairness perceptions, particularly toward AI, play a key role in maintaining employee engagement and cooperation, even in the face of unfavorable outcomes. Lastly, raising awareness about AI's limitations and potential biases is essential. Educating stakeholders equips them to critically and fairly evaluate AI-driven decisions, fostering informed trust and confidence in the technology.

Practical recommendations for organizations

To integrate AI effectively into decision-making, organizations must prioritize transparency by clearly communicating how AI systems operate and make decisions, providing stakeholders with insights that build trust and mitigate skepticism. A hybrid approach that combines AI's objectivity with human judgment can address both procedural and interpersonal fairness, ensuring a balanced decision-making process.

Regular audits and updates of AI systems are essential to identify and rectify biases, promoting equitable and data-driven outcomes. Additionally, developing AI literacy programs can empower employees to understand the technology's capabilities and limitations, fostering informed trust in its use. Even when decisions are primarily AI-driven, empathetic communication from human leaders remains crucial to maintaining trust and engagement, emphasizing clarity and sensitivity in delivering outcomes.