Topic: property prediction

This MRS workshop, developed with the Japan Society of Applied Physics, examines how AI can advance autonomous materials research beyond property prediction. Sessions connect physics-informed models, fundamental physical understanding, experimental validation and human–AI collaboration. Particular attention goes to data-efficient learning where experimental datasets are small or difficult to obtain. The online programme runs 4 November, 10:00–13:00, and 5 November, 19:00–22:00, US Eastern Time (EST). Registered participants receive session recordings. Registration is USD 100 regular, USD 75 fo

Qichao Hu, founder of Molecular Universe and CEO of SES AI, discusses the difference between building AI research tools and delivering new materials. Using battery development as an example, he traces hypothesis generation, candidate formulation, property prediction, experimental ranking, and autonomous synthesis and testing. The conversation examines the difficult connection between molecular properties and device performance, especially the solid-electrolyte interface, and why reliable experimental data and simulation calibration remain central obstacles. Hosted by Ravid Shwartz Ziv and Alle