Anti-LLM Sentiment Considered Harmful

I can resonate with the OP. Learning about AI tools, capabilities, limitations and recent developments has been a hobby of mine for the past few months. I am wholly convinced that in the hands of a good engineer, AI coding agents can massively increase productivity, while at the same time actually increasing the quality of the work. The main reason for that is having much more time left after solving the problem to actually make the solution nice and tested.

There’s indeed a negative sentiment on this forum, expressed by some vocal people. Don’t let it bring you down though. As with any new technology, there will be dissenters. There are also still plenty of programmers who code on effectively a type writer, as opposed to an editor with a language server, multi cursor, extensions and whatnot.

I think the best thing we can do is show really cool and beautiful stuff and say how LLMs helped us write it faster and how it helped improve code quality.

See this PR for example: Making sure you're not a bot!

I used opencode (with one of its free models) to speed up the development of that PR, and I’m very happy with the result. The goal of the PR was to improve typing and to make it easier to run/test the script, while having the same functionality. That makes it very easy to argue that the quality of code strictly improved.

I could have done the same without LLMs, because I have quite some expertise in both Nix and Python. However, I would not have been able to go from never having seen that code before, to a PR in just a single day, and that is the power of the LLM.