Paper Title
Building A Smarter Future: AI-Powered Risk Insights for Construction

Abstract
The construction industry is inherently complex, involving a wide range of stakeholders with diverse goals, which often leads to conflicts and disputes. These challenges are compounded by the numerous uncertainties in construction projects, including schedule delays, cost overruns, and disruptions, which can result in significant productivity losses and economic strain. Effective risk management is crucial to mitigating these issues, yet traditional methods often rely on subjective, manual processes that lack efficiency and consistency. Additionally, existing literature highlights a gap regarding the automation of risk registers, with risk registers often being underutilized despite the recurring nature of risks across projects. To address this gap, this paper developed an AI-driven model to help project professionals uncover hidden patterns within risk registers. The Multiclass classification machine learning model programmed using R software, using a dataset of 326 risk records collected from a contracting company in Egypt. The dataset included features related to risk identification and assessment. The model, which combines Natural Language Processing (NLP) and Gradient Boosting Algorithms, is trained to provide risk response recommendations. The dataset was split into 80% for training and 20% for testing. Gradient Boosting model performed better than Random Forest, achieving accuracy of 96%. This paper holds significant potential for enhancing risk management practices by automating risk response decisions, thereby improving project outcomes and overall efficiency in the construction industry.