AI model gains 3D crystal understanding for materials discovery

MIT researchers have created a multimodal AI model that combines 3D atomic structures with natural language to better understand crystalline materials, outperforming text-only models in identifying promising designs. The system generates rich descriptions linking structure, properties, and applications. Published in Nature Communications, the model uses contrastive learning to align geometric structures with text descriptions. When asked to find solid-state lithium-ion battery electrolytes, its top results matched independently identified candidates. The model also learned geometric representations that text-only models could not. Led by Ju Li and Elsa Olivetti, the team trained the model on AI-generated narratives rather than papers alone, improving function prediction. They compare the approach to AlphaFold, hoping it will accelerate materials science across many crystal types.