AI contributions

[Richard Kimberly Heck]

On 5/8/26 3:14 AM, José Matos wrote:
> On Wed, 2026-05-06 at 23:04 +0200, Jean-Marc Lasgouttes via lyx-devel wrote:
>> I would not object if people who know what they do use AI for such small
>> changes.
>>
>> JMarc
> In my own opinion the question is more nuanced than that.
>
> What does qualify to be a small change? What if there is a set of small changes that when considered together make a great change?
>
> Regarding us, as a project, the issue about AI can come about the issue of the training of the model (who trusts the trustees?).
>
> Another important issue for us is the maintenance of the code. Here is here the slop affects us. My main point, regarding the objection to the "small changes argument", is that we should not trust a huge code change, at once, regardless whether it comes from AI or purely from humans (or other sentient beings). :-)
>
> FWIW I think that we in agreement here. :-)

To my mind, the key difference is between existing contributors and 
submissions from people we do not know, and who have no track record. 
I'm less worried about large vs small. If Pavel wanted to us Claude to 
help re-write large segments of the comparison code, I would trust him 
to make sure that the changes were actually sensible, and trust the rest 
of us to go over the code as well. If some random person submits a bug 
patch, then I'm going to review it carefully whether they used AI or 
not. If it's huge, then of course I'll be more skeptical, and more 
careful, but that's where the size difference matters.

Riki