“Takes on ‘Alignment Faking in Large Language Models’” by Joe Carlsmith | LessWrong (30+ Karma) | Podwise