Damo Academy unveils an AI agent able to discover superconductors (www.scmp.com)

Alibaba’s Damo Academy has introduced Elements Claw, an innovative AI agent that has successfully identified four new superconductors, marking a significant advancement in material science. This breakthrough is pivotal because discovering new superconductors has traditionally been an arduous process characterized by trial and error, as the scientific community lacks a comprehensive theoretical understanding of superconductivity. Historically, only around 2,000 superconducting materials have been cataloged in the SuperCon database, highlighting the need for tools that can accelerate this discovery process. Elements Claw is built on an advanced one-billion-parameter foundation model, trained on an extensive dataset of 125 million molecular and crystal structures. It efficiently screened 2.4 million stable crystal structures within just 28 hours using GPU computing power, identifying approximately 68,000 candidates with superconducting potential. This targeted approach not only expedites the identification of new materials but also promises to enhance research capabilities in the field, potentially leading to revolutionary applications in energy, computing, and advanced technology.