Thanks Amogh also Nicolas for the discussion and comments - I responded to
your questions and did some adjustments, that hopefully might be close to
consensus.

Everyone - feel free to still comment, I am watching and trying to respond
and explain or find ways to address them. I already got plenty of
approvals, so I plan to merge it tomorrow and send "LAZY CONSENSUS" on the
merged version. We can always change it later/revert if the consensu will
not be reached, but from the comments I gather that we all share the same
sentiment.
Looking at many recent PRs - that it is badly needed as apparently some
contributors have very little idea what expectations we have and that
submitting AI generated PR without reviewing it is very bad - for us and
for them. This will hopefully help to educate them.

J,


On Thu, Jan 8, 2026 at 6:00 AM Amogh Desai <[email protected]> wrote:

> I am +1 on this spirit as well and am particularly impressed with how spark
> did it
> earlier in the cycle.
>
> Reviewed the PR template and have a few improvement suggestions over
> there.
>
> Thanks & Regards,
> Amogh Desai
>
>
> On Wed, Jan 7, 2026 at 4:19 PM Jarek Potiuk <[email protected]> wrote:
>
> > I experimented a bit with the PR template - and I came up with this
> > https://github.com/apache/airflow/pull/60158#issuecomment-3718296162
> >
> > Unfortunately we cannot use the .yml format as we do for issues and we
> > cannot make any fields "mandatory" here.
> >
> > I think it's prominent enough and if someone does not check it but uses
> AI
> > we can easily call-out on it.
> >
> > Any comments and suggestions are welcome in the PR of mine.
> >
> > J.
> >
> > On Wed, Jan 7, 2026 at 12:54 AM Ferruzzi, Dennis <[email protected]>
> > wrote:
> >
> > > I like it in concept.   To reiterate, it isn't the use of the
> generative
> > > tool, it's the dependence on it that is the issue.  I think we can add
> > > checkboxes and required fields all we want, but people never read them.
> > > The best we do is add it so we can point to it later, I guess.  Thanks
> > for
> > > writing that up, Jarek.
> > >
> > >
> > >
> > >  - ferruzzi
> > >
> > >
> > > ________________________________
> > > From: Zach Gottesman via dev <[email protected]>
> > > Sent: Tuesday, January 6, 2026 12:53 PM
> > > To: [email protected]
> > > Cc: Zach Gottesman
> > > Subject: RE: [EXT] [PROPOSAL] Gen-AI guidelines in contrib docs
> > >
> > > CAUTION: This email originated from outside of the organization. Do not
> > > click links or open attachments unless you can confirm the sender and
> > know
> > > the content is safe.
> > >
> > >
> > >
> > > AVERTISSEMENT: Ce courrier électronique provient d’un expéditeur
> externe.
> > > Ne cliquez sur aucun lien et n’ouvrez aucune pièce jointe si vous ne
> > pouvez
> > > pas confirmer l’identité de l’expéditeur et si vous n’êtes pas certain
> > que
> > > le contenu ne présente aucun risque.
> > >
> > >
> > >
> > > +1 to mandatory PR description field, as someone who spent time
> reviewing
> > > one of the aforementioned PRs without realizing it was AI-generated
> > >
> > > On Tue, Jan 6, 2026 at 2:58 PM Jens Scheffler <[email protected]>
> > wrote:
> > >
> > > > Thanks Jarek for starting the proposal. As I approved also in the
> > > > devlist my signal, I am +1 on it.
> > > >
> > > > As like in Spark (did not know about this, they did it in 2023!) I'm
> > > > also for adding similar to PR template. Then nobody could claim they
> > > > have not seen / read it. Not all read the contributions docs.
> > > >
> > > > On 1/6/26 20:44, Jarek Potiuk wrote:
> > > > > Question raised in a PR - adding here for better visibility:
> > > > >
> > > > > Apache Spark has a PR template item about the usage of AI added in
> > > > > apache/spark#42469
> > > > >
> > > > > Shall we also add a note into
> > > > >
> > > >
> > >
> >
> https://github.com/apache/airflow/blob/main/.github/PULL_REQUEST_TEMPLATE.md
> > > > > ?
> > > > >
> > > > > Here is the content of PULL_REQUEST_TEMPLATE of Spark - we can
> likely
> > > > also
> > > > > make it optional or mandatory checkbox in the PR ?
> > > > >
> > > > > WDYT?
> > > > >
> > > > > There is a bit of friction if we make it a mandatory field to fill,
> > but
> > > > > maybe it's worth it ?
> > > > >
> > > > > J.
> > > > >
> > > > >
> > > > > ### Was this patch authored or co-authored using generative AI
> > tooling?
> > > > > <!--
> > > > > If generative AI tooling has been used in the process of authoring
> > this
> > > > > patch, please include the
> > > > > phrase: 'Generated-by: ' followed by the name of the tool and its
> > > > version.
> > > > > If no, write 'No'.
> > > > > Please refer to the [ASF Generative Tooling Guidance](
> > > > > https://www.apache.org/legal/generative-tooling.html) for details.
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > On Tue, Jan 6, 2026 at 1:24 PM Aritra Basu <
> [email protected]
> > >
> > > > wrote:
> > > > >
> > > > >> Overall I'm for this, was about to add some comments but unable to
> > do
> > > > it ,
> > > > >> getting some errors. Will add once I get home. Mostly in ways of
> > > adding
> > > > >> some more lines to hammer home the cost of these spammy prs and
> > > updating
> > > > >> some sentence structures. But fully onboard with the spirit of it.
> > > > >>
> > > > >> --
> > > > >> Regards,
> > > > >> Aritra Basu
> > > > >>
> > > > >> On Tue, 6 Jan 2026, 3:42 pm Jarek Potiuk, <[email protected]>
> wrote:
> > > > >>
> > > > >>> Hello here.
> > > > >>>
> > > > >>> We have recently - like almost everyone else - started to receive
> > > some
> > > > >>> Gen-AI generated PRs that are creating some distractions -
> recently
> > > we
> > > > >>> closed 25(!) PRS of a contributor that was clearly doing PRs
> > without
> > > > >>> understanding what their AI proposed, without review or even a
> > touch
> > > of
> > > > >>> understanding what they do:
> > > > >>>
> > > > >>>
> > > > >>
> > > >
> > >
> >
> https://github.com/apache/airflow/issues?q=is%3Apr%20author%3A%22Arunodoy18%22
> > > > >>>
> > > > >>> Some of those PRs looked "plausible" but either tests were
> > completely
> > > > not
> > > > >>> working or the changes themselves were inconsequential.
> > > > >>>
> > > > >>> We discussed it in private@ and I think it's a good idea to add
> > > clear
> > > > >>> guidelines on how to use Gen AI for contributions, point out bad
> > > > >> behaviours
> > > > >>> and make it very clear that similar usages of Gen AI will not be
> > > > >> accepted.
> > > > >>> We should be clear about expectations we have towards such PRs -
> > > while
> > > > at
> > > > >>> the same acknowledging that it's perfectly fine to use AI as long
> > as
> > > > our
> > > > >>> expectations are met.
> > > > >>>
> > > > >>> I also added one thing that is important - it seems that people
> do
> > > such
> > > > >> PRs
> > > > >>> partially because they want to boost their reputation, but as the
> > > > example
> > > > >>> of the contributor that had 25 closed PRs with a maintainer
> saying
> > > "you
> > > > >> are
> > > > >>> doing it wrong, stop" - is ALL BUT boosting reputation - it's a
> > clear
> > > > >> path
> > > > >>> to being a) ignored by everyone b) reported to Github as scammer
> > and
> > > > >>> getting your account shutdown.
> > > > >>>
> > > > >>> I proposed a PR https://github.com/apache/airflow/pull/60158
> and I
> > > > >> welcome
> > > > >>> any comments - this might be a bit sensitive thing, so it's worth
> > to
> > > > have
> > > > >>> more people comment and make sure the bias of single person and
> > > > cultural
> > > > >>> differences will not make it seem too harsh or somewhat drive out
> > the
> > > > >> valid
> > > > >>> contributions.
> > > > >>>
> > > > >>> I do not think we need some specific voting on it, but once we
> give
> > > it
> > > > a
> > > > >>> few days of discussions and give people a chance to look at it -
> i
> > > will
> > > > >>> merge it and send a LAZY CONSENSUS here - because I think we
> record
> > > it
> > > > >> as a
> > > > >>> community approach that we all consent with.
> > > > >>>
> > > > >>> Particularly *Arunodoy18* - if you are watching it and have
> > something
> > > > to
> > > > >>> add in the defense of your PRs - maybe we misunderstood the
> > behaviour
> > > > and
> > > > >>> intentions of yours and maybe you have some other perspective -
> > this
> > > is
> > > > >> the
> > > > >>> right time for you to step up and explain.
> > > > >>>
> > > > >>> J.
> > > > >>>
> > > >
> > > > ---------------------------------------------------------------------
> > > > To unsubscribe, e-mail: [email protected]
> > > > For additional commands, e-mail: [email protected]
> > > >
> > > >
> > >
> >
>

Reply via email to