T. Mustard, T. Sours, A. Singh, O. Allam, M. Cormier, A. Xiao
SandboxAQ,
United States
Keywords: materials science, artificial intelligence (AI), machine learning (ML), large quantitative models (LQMs), accelerated materials discovery, computational design, experimental optimization
Summary:
The discovery of new materials has long been a time-consuming and resource-intensive process, often taking over 10 years to bring from laboratory development to commercialization. Experimental R&D has historically dominated the timeline, but recent advances in computational methods have significantly accelerated this process, with the integration of machine learning (ML) models now enabling AI-guided design of experiments that minimize both computational and experimental cycles. We will explore the exciting potential of large quantitative models (LQMs) to transform materials science, revolutionizing the R&D cycle for new material discovery by leveraging ML algorithms to identify optimal conditions and parameter spaces with unprecedented speed and accuracy. This accelerated approach enables rapid exploration of vast design space, allowing for more efficient testing of hypotheses, automated prediction of material properties that reduces reliance on iterative experimentation, and data-driven optimization of experimental protocols that minimizes trial-and-error methodologies. By harnessing the power of LQMs in materials science, we can accelerate breakthroughs and improve our understanding of complex phenomena; this presentation will discuss recent successes, highlight emerging trends, and provide insights into future directions for this rapidly evolving field.