On 6/18/26 18:50, Simon Tournier wrote:
Hi Ludo,
On Thu, 18 Jun 2026 at 17:30, Ludovic Courtès <[email protected]> wrote:
The version currently checked in at
<https://codeberg.org/guix/guix-consensus-documents/pulls/13> reads
“will not use nor encourage use of genAI to author code or packages, to
review code, […]”.
So it’s not “any use”, it’s the specific uses that are listed, and for
project-related activities.
This doesn’t answer to my concern.
Is it “any use” to author code or packages, to review code, […]?
I’m sorry but I cannot do better than providing a real-world example [1]
with a concrete use-case where the meaning of “use” remains ambiguous
with the current wording.
Do you consider that this use of ChatGPT
<https://chatgpt.com/share/69844b1c-0a10-8011-8308-333ed22e3ed2>
that leads to this author code or package
<https://codeberg.org/guix/guix/pulls/1623>,
would it be a breach of Rule #1?
Just for the record (since I performed the linked experiment): while
this is a good example to discuss Simon's point, the example is quite
trivial and does not represent my envisioned use of LLMs; we don't need
an LLM for this, but that's beside this particular question.
Also, no-one was confused, and I enjoyed becoming better at my craft.
Nevertheless, I'd approach it differently now (almost a year later).
PS: And you cannot escape by applying Rule #2 about some “not legally
significant” code since I’m not contributor but I feel part of “the
project” as someone with write access to one Guix repository.
1: Re: Can a project accept LLMs' code contribution and remain free software?
Hugo Buddelmeijer via "Development of GNU Guix and the GNU System distribution."
<[email protected]>
Thu, 05 Feb 2026 09:52:46 +0100
id:[email protected]
https://lists.gnu.org/archive/html/guix-devel/2026-02
https://yhetil.org/guix/[email protected]
