For millennia, discovering new classes of materials has depended on intuition, experimentation, and often serendipity. However, recent technological advancements have established a more systematic, efficient and robust framework for materials design and discovery. Lately, there has been a surge in studies utilizing artificial intelligence (AI) to identify anomalies, predict stable compounds, material compositions and forecast effective properties. Such computational screening can help avoid unnecessary expenses, save time, and conserve resources during the initial stages of finding optimal solutions. With this motivation, the current review article consolidates research on the broad area of materials informatics, wherein the primary objective is to systematically highlight how the emerging AI frameworks are increasingly being exploited in materials design and discovery. It chronologically discusses online materials databases, data-driven design and optimization, machine learning (ML) and deep learning (DL) frameworks, computer vision-based approaches, natural language processing (NLP) in materials science, and transformer-based foundation models, along with generative and agentic intelligence. This review aims to present cutting-edge AI-driven frameworks in the context of materials science and engineering, providing the research community with critical insights into the progress, caveats and direction of research in this field.
Gupta et al. (Wed,) studied this question.