By now it is clear that ChatGPT and other LLMs have been oversold, thanks to the powerful media presence of their creators and the wishful thinking of audiences. Many of the claimed miracles of such systems have been disproven. Here, just another such research that trashes bold non-scientific claims. This doesn't mean LLMs are not useful, just that the term AI grows the wrong expectations in the human mind
https://arxiv.org/abs/2304.15004