On 10/16/26 23:18, Daniel Henrique Barboza wrote:
Hi,
Not sure if this belongs to the changes made in these patches but can
we establish some "rules of engagement" in the mailing list
discussions w.r.t AI usage in replies? I know that patch 3 touches
on it a little bit but I'm talking mostly about mailing list email.
This is currently handled in both the policy and AGENTS.md. Cover
letters, commit messages, responses to reviewer comments must be written
by people.
we have cases where the
developer keeps sending the same stuff over and over again, and we're
now wondering whether someone is being cheeky and ignoring feedback
or if the developer didn't fully understand the feedback, and I bet
it's the latter.
I'll assume that we don't want to forbid AI usage in ML replies (might
as well just forbid AI altogether in that case)
No, that's exactly what we want. There must be a clear boundary where
we say "from this point on, you must be participating on your own". For
commit messages it may be okay to start with a draft provided by the
LLM, but in-thread replies done with AI are both disrespectful and a
waste of time.
Paolo
