One approach to AI contributions

I recently wrote about taking a careful approach to AI-generated content in communication channels. But what about AI-generated code contributions? There’s no generally-accepted policy for handling those in open source projects, even as many of the tools projects use roll out AI coding assistants. For that matter, there’s not a generally-accepted position on the copyright implications of large language models.

That’s why I like the Apache Software Foundation’s (ASF) guidance on generative tooling. It carefully avoids taking a strong stance on any of the questions that the broader community is still grappling with. The essence of the guidance, to my understanding, is “you can submit AI-generated content if you can promise it doesn’t violate anyone’s copyrights.” I expect the practical effect will be that no AI-generated content will be acceptable, at least in the near term, but it leaves the door open.

All open source communities will need to adopt an AI contribution policy at some point, and ASF’s example is a good one to follow for now. There are a lot of open questions about the legal, ethical, and practical aspects of AI-generated content. We’ll probably be arguing about them for years to come. But your contributors will be trying to submit AI-generated content now.

This post’s featured photo by Mohamed Nohassi on Unsplash.