See my reply further on the thread. Basically the large language models/LLMs are only going to rearrange and regurgitate existing solutions and documentation. If you ask it for programming solutions that aren't trivial, most of the code is wrong, and takes more time to debug or clean up than just writing it from scratch.