We sure as hell could do with some more "common sense" at layers above the AI. Maybe we could avoid this parade of absurdity where people use LLMs expecting them to think.