[Vanclief]

While I like the self reflection from this article, I don't think his methodology adds up (pun intended). First there are two main axis where LLMs can make you more productive: speed & code quality. I think everyone is obsessed about the first one, but its less relevant.

My personal hypothesis is that when using LLMs, you are only faster if you would be doing things like boilerplate code. For the rest, LLMs don't really make you faster but can make your code quality higher, which means better implementation and caching bugs earlier. I am a big fan of giving the diff of a commit to an LLM that has a file MCP so he can search for files in the repo and having it point any mistakes I have made.

[ksenzee]

This doesn’t match my experience. I needed a particularly boilerplate module the other day, for a test implementation of an API, so I asked Gemini to magic one up. It was fairly solid code; I’d have been impressed if it had come from a junior engineer. Unfortunately it had a hard-to-spot defect (an indentation error in an annotation, which the IDE didn’t catch on paste), and by the time I had finished tracking down the issue, I could have written the module myself. That doesn’t seem to me like a code quality improvement.

[malfist]

I don't know what world you're living in, but quality code isn't a forte of ai

[Chinjut]

For what it's worth, this response is explicitly anticipated in the article.