A tip to spy ChatGPT built-in system Prompt

How ChatGPT works under the hood

As I mentioned in my previous story: My Stories are Detected by Medium as AI Generated Content since May,

I write my original story written in my mother language, Chinese, and then translate them into English using ChatGPT.

Unfortunately the translated version, since May, is probably considered as AIGC by Medium :(

So nobody could see my new stories after May except my followers.

Below is a screenshot that I paste one of my translated stories into GPTZero and the scan result shows it has 72% probability AI generated. Frustrated.

Now I am looking into possibilities, if I could use some optimized ChatGPT Prompt, so that the translated text, has less possibility to be marked as AIGC by tools.

Do you think this idea is feasible?

In fact there are many GPT applications in OpenAI’s GPTStore with name containing “Humanize”.

Apparently, the purpose of those GPTs, is to “optimize” the content generated by AI, and remove the AI flavor that reads very boringly.

Of course, could the humanized content generated by those GPTs escape from being detected by AI Scan tool? It depends, and that’s another story which is worth…