The complaints about LLMs on open-source projects I believe are well documented by now. Slop PRs, automated issue spam, unmanageable review backlogs solving issues that aren’t really problems. But I’m interested in what happens when you use LLMs on the flip side of community driven, open source projects. What happens when the maintainers just become the conduits of LLMs.
Interacting with maintainers could feel less like collaborating with a human peer and more like a sterile back and forth with a bot. You open a thoughtful issue or spend time drafting a nuanced PR and receive a reply within minutes: “You are absolutely right! Thank you for bringing this to our attention. We highly value your contribution and will look into this as soon as possible.” It’s polite. It’s grammatically flawless. And it is also completely devoid of life.
People aren’t stupid; we can easily identify AI-generated writing. The distinct lack of a personal voice, predictable formatting (c’mon at least remove your em-dashes!), and the superficial enthusiasm stick out like a sore thumb. If maintainers become mere conduits for LLM outputs, the work might move forward, but the community will be the collateral damage.
My personal opinion is we should not replace authentic human interaction with generative corporate cheer. It leads to a total lack of real engagement. An LLM can summarise your code, but it doesn’t appreciate your work. Getting an automated pat on the back feels less like a validation of your effort and more like being brushed off.
This creates a far less welcoming experience. Ironically, the hyper-polite tone of AI often feels incredibly cold. New contributors don’t want a bot; they want to know a real person looked at their work. Consequently, we lose the ability to build real relationships with real maintainers. Open source has always been a place where relationships form, startup co-founders, and career-defining mentorships are born in GitHub comment sections.
We could be left with a community void of individual personalities. The best projects reflect the quirks, humour, and unique perspectives of their creators and maintainers. When everyone uses the same LLM to write comments, every piece of feedback and point of view sounds exactly the same. And the bit that annoys me the most is that ultimately, it breeds a lack of respect. If a contributor spent hours debugging an issue, answering them with a two-second ChatGPT prompt says, “Your time is valuable, but mine is too important to actually talk to you.”
Sure maintainers are exhausted, and using AI to manage communication friction is an understandable. But if we automate away the human connection, we aren’t building communities anymore, we’re just hosting code repositories. Short, messy, authentic replies will always beat a flawless, fabricated paragraph.
Leave a Reply