AI language tools are for the tools and to be consumed by the tools. That’s the extent of the “intelligence”
They have their uses; however, you have to leverage their strengths, not their weaknesses.
A computer is not capable of generating truly random numbers, and it will never have this ability. For the same reason, AI will never be able to create original content. Lazy people try to use it to create original content, and that's wrong. Last year, when GPT-3.5 and then 4 came out, YouTube was full of grifters "selling shovels" during the AI gold rush. The number of unaccredited experts skyrocketed, and then the "prompt engineers" came along. Well, prompt engineering came and went because LLM models got better at recognizing the input. It is a valid technology; however, like any tool, it should be used accordingly.
For example, it is very useful to correct grammatical errors, read data from scans or images (if your OCR software can't do it), correct logical errors, etc. It's the next step in the evolution of the "bicycle for the mind," if you will. At the end of the day, it's a machine, a complex program that primarily operates based on statistics and complex computations. GPT started showing some of the code used for generating responses, making calculations, etc. People should sometimes click on that little button to see the code so they can stop thinking it's magic.
There is nothing wrong with uploading a document or pasting text into ChatGPT and at the end telling to: "Correct the grammatical errors in the above text." It will do just that, without altering your content. However, as a writer, you still have to proofread what it wrote.
I'm with you there, the only true thing about AI is the "artificial" part, as it is not intelligent at all.