Can Large Language Models Build Causalgia