I asked ChatGPT a question then in a new window asked it to analyze a paragraph I randomly selected from its original answer to determine if it was written by ChatGPT. ChatGPT said the pattern of the text was highly unlikely to have been written by ChatGPT because it lacks the typical conversational tone and linguistic patterns found in ChatGPT-generated text.<p>Then back in the original question window I asked ChatGPT if that quoted paragraph was original generated text by ChatGPT and "it appears to be original text generated by me, ChatGPT".
Is this a question or a statement?<p>All ChatGPT does is predict the most likely next token (with a little sampling on the top most likely tokens).<p>That's it.<p>It doesn't lookup whether it wrote it before, it doesn't use an watermarking tools to check, all it does is predict the next most likely token after the tokens you sent it.<p>It can't even fill in text in the middle, they have to use a special trick where they have a <filltexthere> token.<p>It's been finetuned/trained with text that it's generated since it's been all over the internet since it's launch. However I highly doubt it's been tuned to detect chatgpt text (instead I imagine they'll make it opt to use a tool and use a watermarking tool when they eventually do implement it)<p>You just happened to find two cases were it says yes or no.<p>You should not expect any accuracy from asking it that question.