Thing is, lines of code was never a good metric in software engineering (at least it shouldn’t have been). When people talk about “~% of code” I feel certain dissonance between what they’re measuring and what they’re actually doing. I mean, you certainly want to type less for working code, sure - that’s why programmers develop high level languages. Maybe now we should consider how to code with AI and how well/terrible different approaches are doing. Just saying “99% by AI” doesn’t sound like a good way of saying things are replaced by AI. I mean a lot of things will be replaced by AI, but at least not because of/according to lines of code.