LG AI Research 376

At the recent Materials Research Society (MRS) conference, machine learning and experimental automation emerged as core trends in materials research. In particular, the keynote speech by Professor Alan Aspuru-guzik, a leading figure in AI-based materials development, presented the future of 'Self-Driving' materials development.

Google Deepmind unveiled its achievement of expanding the inorganic materials database 10-fold compared to before, through GNoME. Using an Active Learning-based materials generation method, they secured 2.2 million structures, and in this process discovered over 380,000 new stable materials, including 543 battery-related materials.

In addition, Professor Gerbrand Ceder of UC Berkeley introduced an Autonomous Laboratory called A-Lab. A-Lab is a system that automates the entire process, from AI-based materials synthesis design (Retrosynthesis), to robot-based synthesis, to analysis via X-ray diffraction.

This system rapidly compensates for experimental failures through Active Learning, achieving results of conducting 355 experiments over 17 days and successfully synthesizing 41 materials out of 58 candidates.