Machine Learning and Artificial Intelligence–accelerated Computational Approaches in Materials Science

Material discovery and design are being revolutionised by machine learning and artificial intelligence (AI). Traditional methods are powerful, but they are computationally costly and do not efficiently explore the vast chemical and structural design space. Advances in machine learning have enabled prediction of material properties, accelerated simulations, optimised synthesis conditions, and even inverse design of novel compounds—ranging from supervised, unsupervised, and reinforcement learning to deep neural networks and graph-based architectures. A survey of key methodologies, applications, and related work in machine learning and artificial intelligence-driven computational materials science is presented here. In addition, we outline future directions for physics-based models, autonomous laboratories, and explainable artificial intelligence while discussing the challenges of data quality, transferability, interpretability, and integration with experimental workflows. The pace of innovation in materials science is being accelerated by AI-driven computational approaches.