It's an interesting phenomenon. When I first started using LLMs, I was impressed by its natural language generation capabilities, and thought it could write con

It's an interesting phenomenon. When I first started using LLMs, I was impressed by its natural language generation capabilities, and thought it could write considerably well - using elegant structures, etc and so forth. But after a while those structures became a sort of signature of LLM writing. They repeat the same style way too much, and with enough interactions it becomes grating to read.