Designing Fair AI for Managing Employees in Organizations: A Review, Critique, and Design Agenda

Unfair AI systems directly harm worker performance and retention: The paper documents that when AI systems treat workers unfairly, it leads to decreased worker effort and increased turnover—making fairness not just an ethical concern but a business problem that affects organizational outcomes.

Three distinct types of fairness matter in workplace AI: The research identifies distributive fairness (fair outcomes), procedural fairness (fair processes), and interactional fairness (respectful treatment), suggesting that organizations need to address fairness across multiple dimensions rather than focusing on just one approach.

Current AI deployment in organizations lacks systematic fairness frameworks: Despite widespread use of AI to manage employees, the paper critiques the lack of a theoretical and systematic approach to designing fair systems, indicating that many organizations may be deploying these tools without adequate fairness safeguards in place.

Justice frameworks can guide how to fix unfairness when it occurs: The paper proposes using both retributive justice (accountability and correction) and restorative justice (repairing harm and relationships) as frameworks for addressing instances where AI systems have treated workers unfairly, rather than simply accepting unfair outcomes as inevitable.

Organizations need a comprehensive design agenda for AI fairness: The research calls for applying fairness principles intentionally across different workplace scenarios, suggesting that responsible AI deployment requires upfront planning and design rather than treating fairness as an afterthought.