Strategies for Fair Machine Learning Applications in the South African Construction

Machine learning (ML) transforms the construction industry by enhancing productivity, improving project quality, and mitigating delays and cost overruns. However, concerns about fairness and bias in ML algorithms raise ethical challenges, particularly in South Africa, where historical and socioeconomic factors may exacerbate inequalities. Ensuring equitable ML applications is essential for fostering inclusive and ethical technological advancements in the industry. This study explores strategies to promote fairness in ML-driven construction project delivery. Using a comprehensive literature review and the Delphi method, thirteen experts from academia and industry, specialising in construction digitalisation and AI adoption, contributed to this research. Statistical analyses, including mean scores, medians, and interquartile deviations, were employed to evaluate and rank key strategies for mitigating bias. The findings highlight the importance of active human oversight, continuous model monitoring and retraining, inclusive data collection, strict AI regulations, and increased diversity in IT and policymaking. Furthermore, fostering multidisciplinary collaboration between industry stakeholders, regulatory bodies, and academia is crucial for establishing industry-wide standards and best practices. By implementing these strategies, the South African construction sector can mitigate bias in ML applications, promote transparency and accountability, and ensure the ethical integration of AI technologies. This study provides a foundation for further research and policy development to create fair and responsible AI-driven decision-making in construction.