Re: Thoughts about LLM usage

On Wed, Aug 12, 2026 at 08:01:19AM -0600, Gunnar Wolf wrote:
>Marc Haber dijo [Wed, Aug 12, 2026 at 01:35:14PM +0200]:
>>I'd like to know what it means for you to be "personally 
>>responsible". I have broken key packages more than once without 
>>using an LLM, would I face other consequences if I broke key 
>>packagees with an LLM assisted contribution, any if yes, why?
>
>Well, Debian is after all a volunteer-driven project, where we do not
>have the power to send you to jail to think on what you did wrong. But
>I suppose you broke said key packages out of a mistake, and promptly
>corrected them.
>
>I think you felt personally responsible, and acted as quickly as
>possible to (probably mistranslating from a Spanish expression) “get
>your foot out of the hole”.
>
>It was not just that you found a bug in random code: you found (I have
>also done so, many times) a mistake in a contribution you made. And
>you felt it was necessary for you to fix the error.

It would have been 1:1 identical if I had used AI to craft that 
contribution that caused the breakage.

>I would expect any DD uploading LLM-generated or LLM-assisted
>packaging to be as dilligent with that as with any other packaging
>they do.

That is the case for the way I work, with and without AI, yes. We have 
ballot options that would not leave me that decision any more.

Greetings
Marc