Collective intelligence for AI-assisted chemical synthesis (www.nature.com)

MOSAIC (Multiple Optimized Specialists for AI-assisted Chemical Prediction) has been introduced as a groundbreaking computational framework that leverages the collective knowledge of millions of chemical reaction protocols to assist chemists in synthesizing new compounds. Built upon the Llama-3.1-8B-Instruct architecture and training over 2,498 specialized chemical models, MOSAIC offers reproducible and executable experimental protocols, achieving an impressive 71% success rate in experimental validations. This novel approach not only facilitates the discovery of over 35 new compounds across various domains such as pharmaceuticals and agribusiness but also uncovers new reaction methodologies unexplored during the experts' training. The significance of MOSAIC for the AI/ML community lies in its capacity to address the overwhelming abundance of scientific literature and chemical reactions published annually. As the pace of information growth outstrips traditional methods of knowledge access, MOSAIC's partitioning of vast chemical domains into searchable expert regions delivers a scalable model for AI-assisted discovery. This framework exemplifies how AI can effectively transform and accelerate the process of chemical synthesis, potentially reshaping the landscape of research and development in chemistry by improving the efficiency and reliability of experimental protocols.