Generative AI with prompt engineering in construction: Enhancing predictive slope stability modelling for safe, sustainable, climate-smart mining practices
Muhammad Kamran, Muhammad Faizan, Shuhong Wang, Danial Jahed Armaghani, Panagiotis G. Asteris, Biswajeet Pradhan
Abstract
• GenAI and prompt engineering transform construction and mining industries. • Prompts include key factors influencing slope stability outcomes. • Models show high accuracy in predicting slope stability metrics • Stresses need for transparent and accountable AI in critical tasks. Generative AI (GenAI) and prompt engineering are rapidly advancing in industries such as construction and mining, leading to significant improvements in efficiency, accuracy, and decision-making processes. These technologies are transforming the construction sector by automating tasks and optimizing workflows, thereby enhancing productivity and risk management. This study explores the application of Google’s Gemini AI tool, a notable breakthrough in GenAI, specifically for predictive modeling of slope stability. The Gemini AI tool is utilized within the Python programming language to generate prompts that incorporate key factors influencing slope stability, with the Google Colab interface facilitating prompt generation and testing. Initially, these prompts are employed for data analysis and visualization, followed by their application in both unsupervised and supervised machine learning approaches. The performance evaluation metrics indicate that the integrated approaches, which combine GenAI and prompt engineering, predict slope stability with a high level of accuracy. The model achieved 99% accuracy, with precision, recall, and F 1 -scores ranging from 0.98 to 1.00 for both stable and unstable slope classes. This innovative methodology seeks to advance the implementation of GenAI in civil and mining engineering, offering more precise and efficient solutions for managing slope stability and supporting safe, sustainable, and climate-smart mining operations.