Regardless of the language you’re using, if what you write describes accurately and unambiguously what a machine needs to do, you’re writing code. If it isn’t unambiguous then you’re not a software engineer anymore, you’re the customer.
It’s the difference between a sketch and a technical drawing. One precedes the engineering while the other is the engineering.
You can’t skip the engineering part and you can’t get a machine to do it for you because we haven’t made a machine yet than can think for itself, ask the searching questions and resolve those ambiguities.
An LLM can simulate the process in a way that will fool a non-engineer into believing it’s happening; a human without engineering skills and experience can simulate it even better, but neither produces a satisfactory result.
It’s easy to ignore things you don’t understand, or pretend they’re not important. People do it all the time, in all walks of life, and wonder why things never work out the way they want.
The same is happening right now with software engineering and has been for as long as software has been a thing. LLMs are just the new excuse.
Let your engineers do their jobs to the best of their abilities, and hope that you’ve hired the right ones. You have no idea how to do their job so don’t try to tell them what to use to do it. That way lies disaster. Trust that where LLMs can help them, they’ll use them, and where they can’t, they won’t.
That’s what engineers do. It’s in their DNA.