idk, code was always written only once, but it will always be read many times. Writing code with LLM centric way and just focusing the high level gets hazardous pretty quickly, since somebody actually needs to understand the codebase deep down (or at least the relevant part) when something goes wrong. When you have 100k+ line codebase with multiple contributors with LLM it can very quickly turn into a black box, which nobody understand anything deep down anymore and everybody are more and more incentivized to use LLMs to interact with the thing to get anything real done. Then it’s couple hallucinations away from turning into a fragile mess, where things break unexpected and nobody knows why.
idk, code was always written only once, but it will always be read many times. Writing code with LLM centric way and just focusing the high level gets hazardous pretty quickly, since somebody actually needs to understand the codebase deep down (or at least the relevant part) when something goes wrong. When you have 100k+ line codebase with multiple contributors with LLM it can very quickly turn into a black box, which nobody understand anything deep down anymore and everybody are more and more incentivized to use LLMs to interact with the thing to get anything real done. Then it’s couple hallucinations away from turning into a fragile mess, where things break unexpected and nobody knows why.