Yes. This is probably why decent software developers feel they work harder with AI/LLM. Forcing models to adhere to standards, reuse code, not break things is no easy task. It is very easy to have a lot of bloat. Not saying that LLMs are not incredible.