Millions of new materials discovered with deep learning

In their recent study titled “Autonomous Innovation Robot for Sustainable Material Discovery”, researchers at the University of California, Berkeley, collaborated with teams from Google and other organizations to create a powerful AI tool that can guide materials discovery and experimentation. The AI, named GNoME (Groundbreaking New Material Explorer), is able to search through a vast database of known and predicted crystal structures for new material discoveries. The researchers have identified 380,000 potential candidates using their AI tool and found 736 promising candidate materials from the previously synthesized library. GNoME has been shown to accurately predict potential superconductors and improve existing ones, as well as discover a new type of diamond-like material that could revolutionize energy storage technologies. The discovery method utilizes AI to automate material synthesis and design, allowing researchers to explore previously unexplored materials with less time and resources.