You are forgetting the human mind accounting for this and adding "write this like a kinda dumb high school student". I just did a little test between a copilot essay and the same prompt with "write this like a kinda dumb high school student" and it reads like an essay i would have written.
That's where the humanizers come in. These are solutions that take LLM generated text and make it sound human written to avoid detection.
The principle of training them is quite simple. Take an LLM and reward it for revising text so that it doesn't get detected. Reinforcement learning takes care of the rest for you.
For example “delve” and the em-dash are both a result of the finetuning dataset, not the base LLM.