Not yet human-reviewed
This source record was written automatically by the research pipeline and has not yet been reviewed by the maintainer or integrated into a topic page. Treat it as a provisional summary and verify claims against the original source.
FAIR: A Design Theory for Artificial Intelligence Fairness
Citation. Rai, A., Tian, J., & Xue, L. (2026). FAIR: A Design Theory for Artificial Intelligence Fairness. MIS Quarterly, 50(3), 817–850. https://doi.org/10.25300/MISQ/2026/17971
Summary. Rai, Tian, and Xue develop FAIR (fairness adaptation through AI-augmented responsiveness), a sociotechnical design theory for managing persistent AI-fairness tensions rather than treating fairness as a one-time technical correction. The theory characterizes tensions across fairness principles, goals, and foci, then specifies iterative surfacing and resolving cycles coupled with risk-tiered federated governance. It assigns AI agents monitoring and refinement work while retaining human responsibility for objectives, trade-offs, contextual judgment, and oversight.
Key claims
- AI-fairness tensions are endogenous and persistent because fairness principles, goals, and foci are interdependent across organizational, technical, governance, regulatory, and societal conditions.
- FAIR specifies iterative surfacing and resolving cycles that adapt AI decision artifacts across representation, learning, and calibration layers rather than relying on a one-off intervention.
- Complementary human–AI agent collaboration separates AI monitoring and refinement from human specification of objectives, trade-off adjudication, contextual judgment, and oversight.
- Risk-tiered federated governance can balance standardization with local configuration and human control with AI autonomy across a portfolio of AI decision tasks and artifacts.
Evidence & limitations
- This peer-reviewed theory-and-review article synthesizes AI-fairness perspectives and design-science, paradox, and managing-with-AI literature; it proposes a design theory rather than reporting a new empirical evaluation of FAIR.
- The source is publicly available through the publisher under CC BY 4.0; Crossref and the DOI landing page confirm the title, authors, September 2026 publication, journal venue, and article DOI.