ARTIFICIAL INTELLIGENCE FOR CONTROLLING THE OPERATIONAL EFFICIENCY OF METALLURGICAL ENTERPRISES: A REVIEW OF MODERN APPROACHES, TECHNOLOGIES, AND CHALLENGES

Abstract

This study is dedicated to examining the transformational potential of artificial intelligence (AI) and, in particular, large language models (LLMs) in controlling the operational efficiency of metallurgical enterprises. It examines contemporary approaches, advanced technologies, and key challenges associated with integrating AI into such critical areas as automated manufacturing process management, quality control, predictive maintenance, supply chain management, and workforce management. Based on a comprehensive literature analysis, this article elucidates how LLMs can enhance efficiency, accuracy, and adaptability within the metallurgical industry, and it also identifies gaps in existing research that warrant further investigation. The integration of AI and LLMs in the metallurgical industry faces complex challenges such as the specificity of industry data and terminology, the opacity of “black box” algorithms, insufficient adaptation of models to industrial conditions, high computational demands, and real-time issues. Additionally, there is a need for qualified personnel, cultural changes, and the development of service staff. A comprehensive approach for the successful implementation of AI technologies in the management of the operational activities of a metallurgical enterprise will include targeted research that takes into account practical, economic, and social aspects. The key directions are the development of physics-informed AI with the integration of metallurgical knowledge, the creation of hybrid systems combining LLMs with traditional methods, and the application of unsupervised learning to overcome data scarcity. The priority is to enhance the interpretability of models, integrate with Digital Twins for real-time monitoring, embed safety constraints through predictive control models, and fine-tune universal LLMs for metallurgical applications.