Emerging trends in AI-driven mining hazard management

The management of geotechnical hazards in mining is critical for ensuring operational safety and sustainability, yet its practical implementation often struggles with the inherent geological uncertainty and the dynamic, non-linear nature of mining environments. Artificial intelligence (AI), and specifically machine learning (ML), has emerged as a transformative technology, offering a shift from reactive to proactive, data-driven hazard management. This paper provides a comprehensive review of the evolution and application of AI in mining hazard management, tracing this development from early statistical and ML models to recent advances in deep learning. The review finds that these advanced models have demonstrated significant success in tasks such as rockburst prediction, slope stability analysis, and gas outburst forecasting by identifying complex patterns in large, multi-source datasets from seismic, geological, and operational monitoring. Despite these successes, significant challenges persist: the efficacy of purely data-driven models is often limited by data scarcity, imbalanced datasets, and a lack of physical consistency; the black-box nature of many deep learning models has created a barrier to their interpretability and trustworthiness in safety-critical applications. The future of AI in mining hazard management lies not in isolated algorithms but in integrated, hybrid systems. Physics-informed neural networks (PINNs) and explainable AI (XAI) are identified as critical frontiers. These approaches promise to create models that are not only more accurate and generalisable but also transparent and interpretable to engineering practitioners, bridging the gap between AI's theoretical potential and its practical application in creating safer mining environments.