AI4S framework pioneers intelligent design of metallic glasses: Bridging theory-experiment gap

A new comprehensive review article published in the new journal AI for science (AI4S) offers a strategic roadmap for harnessing artificial intelligence (AI) to revolutionize the discovery and development of metallic glasses (MG).

MG have long held promise for a wide range of industrial applications, from aerospace to biomedicine. However, their rational design remains a formidable challenge due to the inherent complexity of their composition and structure. Traditional experimental trial-and-error methods are slow and inefficient, while computational simulations face limitations in scalability and accuracy. These hurdles impede rapid progress in understanding and optimizing these materials.

The review highlights significant strides made in integrating machine learning (ML) and data-driven pipelines into metallic glass research. It discusses two key paradigms: a foundational “pre-Keplerian” data pipeline that standardizes data collection and processing, and a next-generation, theory-experiment-aligned framework that closes the existing gap between simulation and real-world experimentation. By focusing on modular workflow components, including data standardization, feature engineering, model validation, and iterative experimental cycles, the review provides actionable guidelines for establishing reliable, interpretable, and efficient AI-driven discovery processes.

Crucially, this work emphasizes the importance of developing adaptive, autonomous, and feedback-enabled strategies that can identify high-value material candidates, unravel underlying structure-property relationships, and accelerate the discovery cycle, moving away from slow, fortuitous methods toward principled, intelligent design.

This review underscores that while challenges remain, such as data heterogeneity and the complexity of metallic glasses, the proposed strategic frameworks and ongoing technological advancements present a promising pathway. By adopting these methodologies, researchers and industry can significantly enhance their capability to design novel metallic glasses more efficiently, ultimately unlocking new functionalities and applications.

This work serves as a comprehensive guide for scientists seeking to leverage AI for materials innovation, marking a crucial step toward turning the concept of intelligent, data-driven material design into reality.

The research has been recently published in the online edition of AI for Science, a prominent international journal in the field of interdisciplinary materials science research.

Reference: Huanrong Liu, Shan Zhang, Qingan Li, Bin Xu, Jian Li, Pengfei Guan. Towards Intelligent Design of Metallic Glasses: A Data-Driven Pathway for Closing the Theory-Experiment Loop[J]. AI for Science. DOI: 10.1088/3050-287X/ae424c

---

---