UMD's Julia Mendelsohn is combining linguistics, political science, and AI to explore implicit bias in political discourse while analyzing the societal risks and potential harms of large language models.