Researchers trained an AI to predict which science ideas will succeed. Most AI helps scientists run experiments. This one decides which experiments are worth running. A new paper introduces OpenMOSS. This AI framework learned "scientific taste" from 700,000 pairs of
