Overview

• Google DeepMind's AI model, GNoME, has predicted 2.2 million potential crystal structures, with 380,000 believed to be stable enough for use in future technologies.
• The AI model was trained on data from 69,000 known crystals and can predict potential properties of new materials, aiding scientists in assessing their stability.
• GNoME has generated 52,000 compounds with a similar structure to graphene, and 528 lithium ion conductors, as well as 15 lithium transition-metal oxides that could potentially be used in superconductors and rechargeable batteries.
• In a collaboration with the Lawrence Berkeley National Laboratory, 58 of GNoME's predicted materials were selected for synthesis by a robotic arm, with 41 successfully produced.
• The breakthrough could accelerate the discovery of new materials for applications such as energy storage, solar cells, and superconductor chips.