In just five years, we have advanced LLMs to the point where they can already generate production level code. Yes, sometimes the output contains bloat, but isn’t it our responsibility to ensure it doesn’t? We can easily address that with good prompting, fine tuning and code context with little human intervention. Considering how far we have come in just five years, imagine what these models will be capable of in the future. Im confident that in another five years, LLMs will be so advanced that theywill generate less and less bloated code over time and adapt itself according to our need.