glibc AI policy

[DJ Delorie]

Andrew Pinski <pinskia@gmail.com> writes:
>> If they did not label their work "Assisted-by: LLM" then the contribution
>> must be considered to come from the author.
>
> A violation of this needs to be handled in the same way as a CoC violation.

While I agree in principle, I think we need to be careful about how we
word this and where we draw the line.  I mean, we don't think it's a CoC
violation to not disclose which editor you used, and we consider
misrepresenting your copyright status to be a legal problem not a CoC
problem.

>> We can work to add a section to the "Contribution Checklist" and more
>> generally ask new contributors to review the entire list in a
>> non-judgemental way.
>
> NO. It should very much be in a judgemental way. It should say missing
> disclosing that you used a LLM is a CoC violation fullstop.

I re-read this section of the thread a long time, and I'm not sure what
we're judging or not judging here.  Do we want the reviewer to judge the
code on whether they think it's LLM-generated if the submitter doesn't
say?  Or whether or not LLM was used at all?  Or are we asking the
submitter to not judge our contribution guidelines?  Or judging them on
how well they follow our guidelines?

> I disagree with respect to this. In fact the burden of proof if a
> patch is NOT LLM generated should be on the submittor rather than a
> reviewer mentioning this seems LLM generated.

I think this is consistent with our history, too.  We do typically
disclose when a patch is mechanically generated by a script.

However, proving a negative is difficult.  In this case, as difficult as
proving you have clear copyright status.  Making "prove you're not AI" a
CoC violation opens up the opportunity to use this CoC rule as a weapon
against contributors, and I oppose that.

I invite you to prove your email was not written by an LLM, and consider
how you would react if I chose to shun you until you did.  This is a
worst case scenario, but in today's environment, we have to consider it.