Two AI-MI researchers were on the program at M2S 2026, the flagship international conference on the materials and mechanisms of superconductivity, held in Stuttgart July 19–25.
AI-MI Director Eun-Ah Kim (Cornell University) gave an invited talk, “Interpretable, structure-aware, AI for superconductivity discovery”. The talk presented a structure-aware probabilistic framework that pairs graphlet-based representations of local chemical and structural environments with Gaussian-process learning, predicting critical temperatures with calibrated uncertainty while surfacing physically meaningful descriptors rather than opaque scores. The model recovered trends across known superconducting families, rediscovered superconductivity in nickelates without having been trained on them, and predicted superconductivity in stoichiometric PtPb₃Bi — subsequently confirmed experimentally, with a measured Tᴄ close to the prediction.
AI-MI senior personnel Kin Fai Mak — professor at Cornell and director at the Max Planck Institute for the Structure and Dynamics of Matter in Hamburg — was an invited speaker in the conference’s Two-Dimensional Materials session.
Source: M2S 2026