I am not a native English speaker and that’s why I used AI to answer your questions.
I keep seeing this, or a version of it, on issues and PRs. Reading it over and over again, especially when written by first-time contributors, infuriates me, it triggers an irrational anger in me! I am also not a native English speaker, most of the other participants in this conversation aren’t either.
I am not angry with the people writing that sentence or any variation of it, I am angry about how it negatively impacts our interactions. I didn’t want to sit with this emotion any longer, so here’s a letter to all first-time contributors:
Dear first-time contributor,
As a maintainer I would rather read your „bad“ English answer to my follow-up questions on your issues, or comments on my PR feedback, than an overly long, winding, clearly-AI-generated wall of text that gives me no context and no idea about the human behind the writing. It doesn’t tell me what you know and don’t know, it doesn’t tell me what kind of help you need to be successful.
What makes me so upset is not that I feel disrespected, it’s that it robs us both of the chance to build a proper relationship. It robs you of the opportunity to learn and grow into a community, and it robs me and other maintainers of the opportunity to guide you.
It’s a pointless conversation to have. I could run this whole conversation with my own LLM instead of you and probably get the code or documentation much quicker than getting back and forth between me asking a question and you feeding that question into your LLM and sending me back your answer.
So, please, do us both a favor, let your guard down. Use your own words, use an old-school translation app, or ask your LLM to „copy edit“ your text before sending it to me. Only this way I am able to collaborate with you! This makes open source fun for both of us :-)
Thank you, Severin
PS 1: this is not about not using LLMs at all! Use them while writing the code, use them even to describe your issue or PR. Not always great, but whatever. This is about the point where the conversation happens. When humans interact, skip the machine.
PS 2: Yes, I come from a place of privilege. My native language is close to English, I attended years of English lessons in school and even some extra courses in university. I have no idea how hard it is to learn a language that is alien to your mother tongue and that you have to teach yourself on your own. But I really think there is a difference between you feeding my question to an LLM directly and you writing a response in your words, maybe even in your own language and translating it.
