The examples shown in previous article on same case (ChatGPT producing exact text) makes it look like LLMs only need to see an example once. Or did it see those

The examples shown in previous article on same case (ChatGPT producing exact text) makes it look like LLMs only need to see an example once. Or did it see those examples many times and therefore reproducing it?