https://www.searchenginejournal.com/chatgpt-watermark/475366/#close
This article talks how the chatgpt dectector works to determine whether certain texts is created by Chatgpt or not. The Chatgpt AI developers setup some traits inside it so that it can be dectected. I think this feature is very important because it makes detecting AI generated texts easier. Originally, I thought the text messages generated by Chatgpt was randomly selected in some word-banks that can correctly denote the meaning according to the text generated before and after. However, the random is actually a fake so that a people can find a pattern. This feature is really effective at distinguishing AI generated works even though it has its drawbacks.