I agree with all points here.
- ferruzzi ________________________________ From: Maciej Obuchowski <mobuchow...@apache.org> Sent: Wednesday, November 27, 2024 5:45 AM To: dev@airflow.apache.org Subject: RE: [EXT] [DISCUSS] Publishing articles in Airflow Medium publication and AI generated content 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. Blogs that duplicate content of the documentation, or provide surface level best practices can be a net negative even if they are providing information that's currently valid, since that can change later. In OpenLineage we've seen people coming to Slack or GitHub with problems due to following old, not updated guides in forgotten blogs, rather than updated documentation, and I'm pretty sure the same happens for Airflow. śr., 27 lis 2024 o 12:30 Jarek Potiuk <ja...@potiuk.com> napisał(a): > Hello here, > > TL;DR; We have this very nice Airflow Publication > https://medium.com/apache-airflow - and recently we had some submissions > that were clearly Gen AI generated and provided no value and I think we > need to agree on some general acceptance criteria. > > We had quite a few submissions submitted recently and mostly those articles > were in the form of: > > * Problem description > * What is important > * What are the solutions > * Summary/ Conclusions > > Or another recurring pattern: > > * What is Apache Airflow > * Key Features > * Getting started > * Installation > * initialization > * Core concepts > * Diving deeper > * Best Practices > * Conclusions > > All listed as bullet points describing a very generic way of dealing with > the problems or just extracting stuff from airflow documentation in the > form of bullet points and short paragraphs. Very typical "structure" for > AI-generated content. > > Together with Briana we decided to reject those publications - they were > really not adding any value and iMHO they "increase entropy" of Airflow > knowledge rather than decrease it. > > I thought (after doing it) that it would be great to agree that this is > the right thing to do and generally agree to some very general acceptance > criteria for those publications. > > In the past we generally accepted pretty much all kinds of articles - > articles for beginners, advanced topics after a brief review if the article > did not have any misleading information / hallucinations / bad advice for > the users. Those are submitted by authors who we accepted as writers to the > publications. > > But IMHO accepting such AI-generated content that increases the entropy Is > bad. > > But I think the right approach for anyone who wants to submit an article > that while it's good to use AI for some part of the content and to help to > generate such articles, the end results should be somewhat insightful > and should "decrease the entropy" of Ariflow knowledge rather than > "increase the entropy". > > While this is difficult to judge, and it's more of an arbitrary decision, > maybe we should agree that it is the right thing to do. I am not sure if we > want to have some body or a group of people to decide whether the article > is good to publish - I feel somewhat uncomfortable - even with Briana > together - to make some arbitrary decisions there. > > Would love to hear what you think and how we could make it more of a > community decision. > > J. >
