Related: How can we identify ChatGPT-generated posts?

Some sites have started to tell moderators to ban users using AI to generate content and reputation.

It can be obvious when a user is using ChatGPT to generate answers on Stack Exchange, and moderators may ban users if they start to give a lot of answers without references. But to copy literal AI answers is not the only way to use the AI.

There are tools like this one that supposedly detect if a text has been generated by AI and suggest you to modificate the text until it appears to have been written by a human. I wonder if these detectors work properly, as the battle against engine use in chess is a difficult one, and language is much more complex than chess.

So, what if a user gets concepts for questions or answers from the AI but modifies the text slightly? How can Stack Exchange sites that follow Stack Overflow policy and ban AI use detect it? To use the AI itself is not an option (at least ChatGPT), as it fails to detect its own text:

Will the community begin a battle against the use of AI in Stack Exchange, as in chess, or is it a lost cause due to the complexity of the programming challenge (where AI would inevitably win and "cheaters" could not be detected)?