This is a good article that explains why AI isn't likely to replace programmers ever.
It could also be extrapolated that most other high-level creative endeavors will also not likely be replaced by AI either. From the article "Automatically discovering formal (i.e., unambiguous) solutions to novel problems can never be automated."
While I don't know enough about the formality of things like languages, which the current state of the art AI (LLM) is presumably excellent at, it suffers from the last mile issue. Much like the current state of the art for automated driving, you can translate the bulk of the text with AI, but making any text so that it's readable needs the touch of humans to adjust for colloquialisms and other minor speed-bumps.
Similarly, I would never trust any automated driving systems except on clear sunny days with no unambiguous situations, which is almost never.
-- Addendum
Every new breakthrough in AI, as LLMs have been, has been accompanied by proclamations that AGI (Artificial General Intelligence) is just around the corner. It's not true today and wasn't true the last time either. If you have any doubt about how complex the task of replacing any neural capability is, just look at the recent news on mapping a cubic millimeter of brain https://www.nature.com/articles/d41.... We have just in the last few years been able to map a complete brain of a nematode, a very simple flatworm.
This understanding of natural neural systems are at the limits of our capability. AGI is a fantasy.