Pwlot comments on A multi-disciplinary view on AI safety research