Algorithmic fairness in AI-driven project management for post-conflict and developing regions: a critical review and framework for context-aware AI

Artificial Intelligence (AI) is poised to revolutionize construction project management, offering unprecedented gains in efficiency, scheduling, and risk assessment. However, deploying AI uncritically in the data-poor and socially fragile environments of post-conflict and developing regions creates profound ethical risks. Standard technical solutions for algorithmic fairness, which require large, clean datasets, are fundamentally misaligned with these realities. This misalignment risks automating historical inequalities and eroding public trust in reconstruction efforts. This critical review synthesizes a multidisciplinary body of literature to map the intersection of AI applications in construction with the specific ethical challenges of these vulnerable settings, including pre-existing, technical, and emergent biases. Synthesizing these findings, we propose a "Context- Aware AI Fairness Framework," a holistic, lifecycle approach structured around four pillars: (1) Foundational Scoping & Participatory Design, (2) Data Governance & Bias-Aware Data Management, (3) Contextualized Model Development & Mitigation, and (4) Human-in-the-Loop Deployment & Continuous Monitoring. The paper concludes by arguing that the prevailing fixation on "big data" is a critical limitation and calls for a new research direction focused on developing robust AI systems that can effectively reason with "small data" and integrate rich qualitative inputs, thereby ensuring that AI serves as a tool for equitable and sustainable development rather than a driver of a new digital divide.