Focuses on research that actively explores methods and strategies to ensure language models' outputs align with human values, ethics, and intentions, constituting a significant portion of the paper's content.