I'm getting tired of the hype around LLMs being able to replace human thought and expertise - they're great tools, but let's not forget they're only as good as the data they're trained on.