I actually don’t think I am factually wrong, as you say, but...

I actually don’t think I am factually wrong, as you say, but maybe you’re right because all I have is 20 years experience as a software engineer and personal research and usage of AI as a coding tool, as opposed to your hype generating surveys from a company with a financial interest in AI coding tools.

When I say “some cases” I don’t mean only a small number of people use it, I mean it is only currently useful for a small portion of software engineering tasks.

Using it in your workflow is not the same as “it’s amazing” in my opinion.
Saying how many people use it is not the same as saying how useful it is to those people.
That stat also misses other stats like how maintainable and extendable is the code it outputs compared to not using it, I.e. what effect does it have on code quality and future effort.

It does very basic things well that can save time, but it’s objectively bad at doing anything complex, any tasks that have broad context, any meaningful work on large legacy code bases, etc.
It also requires effort to check its correctness, which can be a problem when using it for complex things or for lots of small tasks that aren’t similar to each other (lots of time to check can sometimes outweigh the time saved).
That checking is something that still requires the user to have their own skills and knowledge.
If you don’t check it or don’t know how to check what it has written you will create garbage software for sure.

It is useful for sure but nowhere near as much as marketing from AI companies would have you believe.