Toward a pledge on « AI »

standing up for human crafting

Note: The context of this personal note is GCD 008 Standing up for human crafting. As Ludo’ proposed: I would like us to view this discussion as if we were sitting in a circle and had one talking stick being passed around, and I refrain to monopolise this talking stick. For the good, the bad and the ugly, I already had enough times the talking stick in my hand.

That’s said, not asking again for the talking stick does not prevent me to sketch on my own paper, still sitting in the circle. Here is the record of my own draft.

For what it is worth, I’ve read all the messages, back and forth, of the lengthy Pull Request discussion also scattered over guix-devel mailing list. What I’m trying to resolve here is to keep unity beyond our diversity, still keeping my understanding of the “origin spirit“ of GCD 008. The three main knots appears to:

Last, I’m grateful that Software Heritage pioneered a thoughtful direction about LLMs on October 2023.



This document is not an official Guix document!

Summary

Guix is a free software project; it is made by people for people, with knowledge sharing and empowerment as some of its core values. This document proposes a clear stance and a policy to support this vision.

Motivation

Free software as a project is fundamentally about humans: it is about producing software that guarantees the freedom of its users, and in particular freedom from the original developers; it is about empowering people, increasing their autonomy, by providing them the means to study and adapt the software to their needs; it is about knowledge sharing and mutual aid.

Guix as free software contributes to an ecosystem of digital commons that embodies decades of human creative effort and that is now used as raw material for stochastic text generators—large language models (LLMs), also referred to as “generative artificial intelligence” or “genAI”—primarily backed by large corporations.

However impressive the automation allowed by genAI in assisting with software development tasks may look, our community believes genAI has an impact that undermines the very foundations of free software and Guix.

The legal and ethical framework governing genAI is complex and rapidly evolving. On the legal side, while copyright law traditionally permits reading by humans, the large-scale extraction of knowledge from code for genAI training enters uncharted territory. This uncertainty is further intensified by ongoing legislative discussions, such as European Union, US Copyright Office and White House, reflecting the regulators’ struggle to keep pace with scientific and technological advancement. On the ethical side, the definition of what “open source AI [model]” means is very fresh, and concrete operational implementation are still at early stages.

Still on the ethical side, many in our community share their concerns about breaching the reciprocity baked into copyleft licenses, already happening with chardet LLM-assisted “rewrite”, or EmDash WordPress reimplementation in TypeScript “under the more permissive MIT license”. Also many in our community share their concerns about fair sharing of natural resources and the habitability of the planet: the ecological footprint of genAI is well documented, going from the energy, water, and materials needed to build the servers used to train models and to service requests to the land usage of those data center, and to the energy and water needed to operate those data centers.

As we strive to preserve this ecosystem of free software for future generations, this document proposes the adoption of a pledge in support of human production by people for people.

Detailed Design

We propose the adoption of a pledge on behalf of the project, with the following goals:

  • Contributing to the public debate on these matters and creating ties with like-minded organizations and grassroots movements.
  • Strengthening support for craftspeople the project interacts with—translators, artists, developers, and so on.
  • Questioning the reasons that make genAI feel necessary for people using Guix, and finding ways to fill the gap.

This pledge is not about claiming a “moral superiority” of the project and its members for not resorting to genAI, nor is it about “virtue signaling”. We are well aware that commercial genAI services perform well on tasks relevant to Guix, first and foremost packaging. There is no doubt that genAI is already being used within the community. Our goal is not to judge what individuals are doing, nor not to point fingers at individuals. Instead, this proposal aims at setting a standard for what we do collectively within the project.

Pledge

We propose the following project commitments:

  1. The project (defined as maintainers, team members, and anyone with write access to a Guix repository, including Weblate, or to Guix resources such as the build farm) will not use or encourage use of genAI. Namely, the use of genAI means not simply pasting answers, code or packages, code reviews, artwork, translations or comments generated by genAI without proper review using human own input.
  2. We kindly ask contributors to respect this choice and apply it equally. We ask to all contributors to never use genAI for directly or automatically interacting with the project.
  3. The project may incorporate contributions assisted by genAI as long as they are not “legally significant” to ensure the contributor has a valid copyright claim on the code. As a rule of thumb, this includes code less than 15-line-long, or package definitions that are evidently not creative, similar to those that guix import and similar tools might produce.
  4. The project kindly asks to not propose contributions assisted by genAI outside packages. Namely contributions assisted by genAI may be accepted only on the gnu/packages sub-hierarchy.
  5. Software where most of the code was authored or co-authored by genAI will not be packaged in Guix. Notable examples of such code include Claude’s C compiler, EmDash, and Neomacs.
  6. The project will keep working to provide people of all levels of experience with the resources to use Guix and to contribute to Guix without feeling the need to resort to genAI:
    • by facilitating access to documentation and improving it to better suit the needs to people with different levels of experience—newcomers discovering free software, people with prior exposure to GNU/Linux, developers;
    • by developing and improving tools that make it easier to get started such as Guix Packager and guix import;
    • by improving diagnostics to make our tools more approachable;
    • by providing communication channels anyone can use to look for help or to offer support, where all and everyone can feel safe to participate;
    • by improving tools for mechanical translation, refactoring, and updates such as guix import, guix style, and guix refresh, so that tedious packaging tasks can be automated in a way that is transparent, deterministic, and maintains the project’s sovereignty.
  7. We acknowledge that the project’s sustainability depends on automation for all the mechanical, labor-intensive tasks such as package updates. We will keep improving hackable tools and services to automate some of the package collection maintenance work.

Publication

The Guix project will publicize this pledge, by referring to it in its contribution guidelines, by making it easy to find, by presenting it in blog posts, and through any other communication deemed appropriate.

Related Work

A number of free software projects have adopted a policy with respect to genAI, revealing different sensibilities and choices. Here are some of those we looked at:

Software package lists:

Cost of Reverting

Should consensus be found via a new GCD, the pledge could be removed or amended.

Drawbacks and Open Issues

This proposal takes a clear stance that not everyone may agree with. This could lead to fragmentation within the Guix community, or within the free software community.

This document is not an official Guix document!



Feel free to say your word using the talking stick!

© 2014-2026 Simon Tournier <simon (at) tournier.info >

(last update: 2026-05-28 Thu 18:33)