Can't this be automated by having an LLM summarize the video's transcribed text into a concise, non-clickbait title?