Study of 150 developers shows AI generated code no harder to maintain long term

The real insight might be that code maintainability is primarily determined by development practices, not origin. If you're skipping tests, avoiding refactoring, not doing code reviews, and blindly accepting generated code - of course it'll be hard to maintain. But that's true whether a human or AI wrote it.

So the generated code is as good as the developer's code? Is this because of great AI, crap developers or poor choice of problem space? Are they seeing generated code that’s outside the training space hold up? IE, without memorized solutions…