I'm worried that AI coding agents might be removing one of the most important parts of the open-source ecosystem: the learning ladder.
Traditionally, becoming an OSS contributor was a gradual process.
You started with documentation fixes, small bugs, or simple improvements. Over time, you learned the codebase, understood the project's conventions, built relationships with maintainers, and eventually earned their trust.
Those small contributions weren't just about getting things fixed. They were how new maintainers were born.
But now, why would a maintainer spend time reviewing a newcomer's PR for a simple bug when they can ask their coding agent to fix it in minutes?
From the maintainer's perspective, this makes perfect sense. Less review overhead, faster fixes, and more control over the implementation.
But what happens when we automate away the tasks that helped people learn?
The problem isn't that AI makes contributing code easier. It's that the opportunities to learn through contributing might become harder to find.
And if newcomers are expected to understand a project deeply enough to identify complex problems and guide agents toward solutions, how are they supposed to develop that understanding in the first place?
Maybe we need to rethink what a "good first issue" looks like in the age of AI.
Because OSS doesn't just need more code. It needs a way to grow its next generation of maintainers.
I'm curious whether other OSS maintainers are already seeing this happen, or whether AI is actually creating new ways for newcomers to get involved.
Do you like what you are reading? Subscribe to receive updates.
Unsubscribe anytime