This paper aims to explore the effects of machine learning biases in the digital tools used within the Nigerian construction industry.
The study adopts a qualitative research design to identify machine learning biases in digital tools and to evaluate their effects on construction project performance. Construction professionals with practical experience in the use of digital technologies and good knowledge of machine learning biases were interviewed online. The data obtained from the interviews were analyzed using ATLAS.ti software.
The study analysis shows data bias, model bias, human bias and sensor bias as the most prevalent biases affecting digital tools in construction. These biases contribute to various challenges in construction project performance, including increased project costs, safety risks, extended timelines, resource waste, project delays, flawed decision-making and reduced work quality.
Though digital tools enhance processes in the construction industry, findings from the study imply that machine learning biases in digital tools and technologies cause inaccuracies that adversely affect construction project performance. This situation inhibits the competitiveness and sustainability of a people-centered, highly litigious and complex construction industry.
This study provides empirical evidence of the effect of machine learning biases on digital tools used in the construction industry. Expanding existing knowledge on machine learning biases can build greater trust in digital tools and maximize their benefits while minimizing unintended consequences. To achieve this, it is essential for stakeholders in the construction industry, including manufacturers and users of digital technologies, to become well-informed about these biases. By working together, they can develop effective strategies to mitigate these issues and ensure the successful implementation of digital tools.
