AI ethics in the workplace: why a list of general principles is not enough
The promise of “responsible AI” may sound reassuring when it appears in a company policy or a tech pitch, but what does it mean for a job applicant who does not know why a screening system rejected them, or for an employee whose productivity is measured by a system that does not explain its logic? A new working paper from the International Labour Organization brings the question back to firmer ground: AI ethics does not start with a nice list of principles, but with rights and work relationships that already exist.
The paper, published on 23 October, examines the link between five widely cited AI ethics principles and the ILO’s normative framework. The principles compared are fairness, beneficence, non-harm, autonomy and explainability. The paper does not propose new law or binding regulation; it offers an exploratory reading of what may happen when these principles enter the actual workplace.
The difference is not only linguistic.The paper explains that AI ethics principles are broad and generic, often focusing on the individual, their privacy and their ability to understand a decision. The ILO framework, by contrast, is directly tied to the world of work and gives greater weight to collective representation, freedom of association, collective bargaining and social-justice conditions. This does not make the two frameworks contradictory, but it prevents reducing worker protection to a vague claim such as “the system is fair.”
The gap becomes clear when systems are used for hiring or management. Explainability matters when a candidate asks what happened in their screening decision, and data governance matters when an organization collects employee information, but the paper also raises other questions: Who participates in setting the system’s rules? How is its impact on the workplace reviewed? What happens when digital monitoring turns into continuous surveillance or into administrative decisions that are hard to contest?
Transparency alone does not settle the dispute.A company may describe how a tool works, yet the question remains about its ability to affect pay, evaluation or access to opportunities. For this reason the paper distinguishes the AI ethics focus on individual privacy, autonomy and explainability from the international labour law focus on the collective and institutional dimension. The conclusion is not that one approach replaces the other, but that AI governance at work needs both.
That is a practical professional point for regional markets. The paper does not claim that any specific hiring system in the Middle East violates a standard, nor does it provide a regional benchmark for tool usage. It does offer a useful lens for anyone building, buying or operating an AI-based system: separate the technical question of the tool’s accuracy from the professional question of its impact on worker rights, grievance mechanisms and participation.
What is worth asking the applicant and the employee?When facing an AI-supported interview, assessment or productivity program, the most useful question is not whether the organization uses the tool, but where it enters the decision-making process. Is it an aid to a human reviewer or does it generate a ranking that determines who moves to the next stage? What data does it rely on? Can a human review the outcome? Who controls the appeal route when the result appears wrong? These questions do not assume bad faith; they shift the discussion from technical awe to accountable scrutiny.
For employers, the paper is a reminder that principle language does not replace the design of clear procedures. If a tool influences hiring, evaluation or monitoring, serious governance requires assigning human responsibility, data rules, appropriate explanations and a way for affected people or their representatives to take part in dialogue about the impact. AI may speed up decisions, but it does not answer the core question: who remains accountable when people’s work is affected?
For readers in the Gulf, Egypt and the Levant, the practical takeaway is simpler than the theoretical debate: before adopting or purchasing a screening or assessment tool for your organization, ask about the appeal pathway and who reviews the decision in person. The paper does not measure the reality of regional markets, but it gives you ready-made questions to put on the HR table in Riyadh, Dubai, Cairo or Doha.