Hi Pjotr and all,

Pjotr Prins <[email protected]> skribis:

> I think the environmental statement is the weakest part of GCD008, or
> at least the most exaggerated.

I disagree, but that’s OK: each one of us values and prioritizes things
differently, and yet we can find common ground.

> Automating package writing in Guix is a boon for *our* projects. It *is*
> saving us significant time with better outputs.

I trust you on this.

One question I have is: What can we already do that would better address
packaging needs like yours?  Is it possible to reiterate what was
achieved with Guix-CRAN¹ for more packages?

¹ Fully automated import of ~20k packages with 95% of successful builds:
  <https://codeberg.org/guix-science/guix-cran>.

> […] I don't like GCD008 because it is a
> loaded gun aimed at our own feet.

I understand why you would see it that way, on the grounds that LLM
would help us automate packaging and refactoring labor that our tooling
does not (not yet) handle.

I’m seeing the other side of this coin.  My concern, as someone who
already spends more time reviewing than writing code, is that letting
“AI” in without regulation would also be a “loaded gun aimed at our own
feet”.

I listed examples to share my perspective, with questions to whoever
who’d rather avoid regulation as proposed by this GCD:

  
https://codeberg.org/guix/guix-consensus-documents/pulls/13#issuecomment-16470989

Applying the proposed pledge would change nothing compared to current
practices; withdrawing it—and thus sending the signal that “AI” use
won’t be regulated—would have social, technical, and juridical effects
that I think will quickly become irreversible.

I think we have to come to an agreement about a policy that takes into
account these two sides of the coin.

Thanks,
Ludo’.

Reply via email to