When Computational Fairness Is Not Enough: Explaining the Gap Between Algorithmic Fairness and Organizational Justice from a Business Ethics Perspective
This conceptual study proposes a business ethics framework to explain why technically fair AI systems may fail to foster perceptions of organizational justice, arguing that factors like transparency, human oversight, and ethical governance are essential for bridging the gap between algorithmic fairness and employee trust.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the modern workplace, a quiet revolution is underway. Companies are increasingly turning to artificial intelligence to make decisions about hiring, promotions, and performance reviews. The promise of these systems is a world of perfect consistency, where every employee is judged by the same strict rules, free from the fatigue or bias that can affect human managers. To ensure these machines do the right thing, computer scientists have spent years developing "algorithmic fairness." This is a technical concept, a way of checking the math to make sure the computer does not systematically favor one group of people over another based on race, gender, or age. It is a necessary step, but it treats fairness as a purely mathematical problem, solvable by better code and cleaner data.
However, the people actually living through these decisions are not computers. They are human beings who judge fairness not just by the final number, but by how they are treated during the process. They care about whether they can understand why a decision was made, whether they have a chance to speak up if something goes wrong, and whether a real person is ultimately responsible for the outcome. This is the realm of "organizational justice," a concept from business ethics that looks at the human experience of fairness. For a long time, experts assumed that if the math was fair, the people would feel it was fair. But a new line of thinking suggests this assumption is dangerously incomplete.
A recent conceptual study by Seyed Jalalaldin Hosseini, a trainer with the Technical and Vocational Training Organization of Iran, challenges the idea that a mathematically perfect algorithm automatically creates a fair workplace. The paper does not present new data from a specific company or a new software tool. Instead, it builds a theoretical framework by weaving together existing research from computer science, business ethics, and organizational psychology. The author argues that there is a significant gap between what the computer calculates as fair and what employees actually feel is just. This gap, which the paper calls the "Ethical Gap," exists because organizations often focus entirely on the technical quality of the algorithm while neglecting the human systems that surround it.
The study suggests that an algorithm can be statistically unbiased and still feel deeply unfair to the people it affects. Imagine a hiring system that perfectly avoids discrimination in its data but operates as a "black box," offering no explanation for why a candidate was rejected and providing no way for them to appeal the decision. In this scenario, the math is clean, but the experience is alienating. The author proposes that for an organization to be truly just, it must do more than fix the code. It must build a system of ethical governance. This includes ensuring transparency so people know how decisions are made, providing clear explanations, maintaining human oversight so a person can intervene, and creating genuine opportunities for employees to voice their concerns.
The paper introduces a specific model to explain how these factors interact. It posits that ethical governance acts as a bridge, or a translation mechanism, that turns a technically fair output into a feeling of organizational justice. Without this bridge, even the most advanced AI can erode trust. The study outlines four possible scenarios to illustrate this. The best outcome is "Aligned Legitimacy," where the math is fair and people feel it is fair. The most common failure, according to the author, is "Legitimacy Deficit," where the math is fair but people feel unjust because the process lacks transparency or accountability. There is also a dangerous scenario called "False Legitimacy," where people feel things are fair simply because they do not understand the hidden bias in the system. Finally, there is "Manifest Injustice," where both the math and the feelings are wrong.
Hosseini's work suggests that the solution is not to abandon artificial intelligence, but to change how we govern it. The paper argues that organizations must stop asking only, "Is the algorithm fair?" and start asking, "Is the organization fair when it governs the algorithm?" This shift moves the responsibility from the code itself to the people and policies managing it. The study highlights that as AI takes on more high-stakes decisions, like firing someone or denying a promotion, the need for these ethical safeguards becomes even more critical. If an organization relies on a machine to make a life-altering decision without offering a human path for review or explanation, it risks losing the trust and commitment of its workforce, regardless of how accurate the machine's calculations are.
Ultimately, the paper concludes that technical accuracy is a necessary condition for fairness, but it is not enough on its own. True organizational justice emerges from the combination of a well-designed algorithm and a well-designed human process. It requires a culture where employees feel respected, informed, and heard. The author suggests that future research should test these ideas in real-world settings to see how different industries and cultures respond to these ethical gaps. For now, the message is clear: building a fair future of work requires more than just better software; it requires a commitment to the human values that software cannot calculate on its own.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.