It's a very interesting and tough challenge to prevent large language models from hallucinating.