Personal Policy for AI driven contributions
Since 2025 things started to change in the software industry, a new tool was in town, the "AI", this new tool is rapidly changing the way people work in software development, enabling people that are not able or don't have time to learn how to code to build software, even at a faster peace of an expert developer, this has also an impact on open source development with a new wave of projects and contributors to existing projects, that come from people with not strictly technical background, or also "traditional" expert developer that contribute without the need to go through all the technical details of a project.
This change is also re-shaping the way people works and interact with each other on open source software and not, but often the AI driven contributions in any of their form, bug reports, feature requests, code contributions, documentation contributions, are increasing the effort needed by the maintainer of a project, this is because they often come with:
- Contributions that do not follow the guidelines
- Contributions that come with really long descriptions, in some case running around with concepts and not reaching the point
- Good sounding reasoning but wrong or not applicable to a specific project
- A style of communication with rapid verbose replies that maintainers struggle to keep up with
These behaviors existed also before the "AI wave", but were less common, for humans is costly to write hundreds of words to a project without understanding it properly, also human contributors most of the time if informed would understand and change their behavior if something was not the right thing to do in a specific project, "llm" generated contribution are limited in context, any explanation/teaching is lost outside the specific context, humans are happy to learn and bring the right behavior also in other contexts or future interactions.
Because of the way "AI"/"llm" produce contributions, many people totally rejects/ban any "AI"/"llm" driven contributions, even though others accept happily and often blindly("AI is smart"), and may also use and overuse this tools by themselves on their own projects, these are all legitimus approaches in these changing times, but I want to go to a different way here is my approach.
My approach to "AI" or "llm" driven contribution is that I will accept in general contribution done with these tools but only from people/organizations that do sponsor me, or I have a consultancy contract with, this is because some work of "llm" handling would be offloaded by to the contributor to me, so to do that work I want to be adequately payed.
You would ask me "how do you recognize the AI contributions", well I will do a "touring test" if the contributions has the right content at the right level coming from a contributor that has a real world need, and it solves/reports a problem that is valid and meaning-full then the contribution will pass, if the contribution "do not pass the touring test", than you need to pay me to "tutor" your "AI".
Obviously this will have some false positive and false negative, but if someone tutored his AI enough for me to not recognize it as an automatically generated contribution, and the contribution do bring value to the projects I work on than is OK to accept it. On the other side if a human user do so little effort to do a contribution that looks like a cheap "AI" maybe he should pay me!
So yes "AI" contributions are more than welcome, you can find the sponsor links on this website.
It's all for now, I hope also other open source software maintainer start to get use this approach to "AI" contribution.
Ko-fi