It seems to me that there's a Hidden Assumption in the suggestion that we ought to use GenAI for code generation: the assumption is that producing code faster is always better. And it's just plain wrong.
The assumption may be partially — maybe even mostly — right in the world of commercial software — software written to make money, gain more customers, cut costs and so on — but it's patently untrue in much of the Open Source world where software gets developed because someone has a problem they want to solve or a way of doing things they want to improve or they're looking for a way to gain status in the community (the ego motivation), and that development is done without any commercial pressure. There's no particular benefit in a shorter time-to-market. There's not usually any competitor to worry about — indeed, the work may well be imitating or replicating some already-existing solution that works.
Remember that Linux itself came about because, as a university student, Mr Torvalds could not afford to buy a commercial Unix.