I don't mind someone trying LLMs to see if they can do better than existing ML solutions.