ChatGPT is not designed to fool us into thinking it’s a human. It produces language with a specific tone & direct references to the fact it is a language model. I am confident that an LLM trained specifically to speak naturally could do it. It still wouldn’t be intelligent, in my view.