I'd love to see more individual researchers openly exploring AI safety from a scientific and humanitarian perspective, rather than just the technical or commercial angles.