The Django UUID Story with Paolo Melchiorre
Published October 26, 2026
This video features Paolo Melchiorre at DjangoCon Europe 2026 in Athens, Greece.
TALK: AI-Assisted Contributions and Maintainer Load by Paolo Melchiorre
AI tools are increasingly used by contributors to read code, explore codebases, and generate changes. In many open source projects, this is already changing how issues are opened and how pull requests are submitted. While AI can help people get started, it also creates new challenges for maintainers. This talk is based on real discussions and concrete examples from the open source community. The talk also looks at how projects are responding. Some add documentation, disclosure rules, or review guidelines. Others start wider discussions about governance, sustainability, and legal risk. These responses show that the issue goes beyond individual pull requests. Instead of giving simple answers, this talk shares the real questions the community is asking today, and helps contributors and maintainers think more clearly about the future role of AI in Django and open source projects.
More information: https://2026.djangocon.eu/speakers/BVRQXW/
Follow DjangoCon Europe on social media:
*X - https://x.com/DjangoConEurope/
*LinkedIn - https://www.linkedin.com/company/djangocon-europe/
*Mastodon - https://fosstodon.org/@djangoconeurope
#django #djangocon #djangoconeurope #djangoconeurope2026
Automatically transcribed, so expect mistakes in names and technical terms.
Speaker 1: So the next talk is uh Paolo Melchiore, if I pronounced it correctly, hopefully I did. Uh he's a Python developer, Django developer, Python software fellow. Many things. I have to read them all. It's uh he does many things, that's uh that's for sure. And uh he's going to talk about uh maintainer uh fatigue or maintainer load then uh you know all these new a AI assisted uh contributions and how to deal with them. So As soon as we're ready, we're ready for Paolo. Let's uh let's give an applause for Paulo and thank you very much for being here and giving this presentation.
Speaker 2: So thank you. Thank you very much for being here and attending this talk and thanks for the organizer to add this talk in the schedule. It's something I'm seeing more and more in our community and I was happy to share with you what I found out. uh on the subject. I'm seeing more and more problem connected to the AI assisted contribution, especially for maintainers and maintaining at the core of our community and we want that they um are happy and our community can be happier.
Speaker 2: I 'm going to talk to you not from the point of view of an expert of both topics, but based on my observation. from my personal experience. It's not a talk about AI alone or against AI. It's only a supervision of what is happening now in our community. And this is, as I said, my point of view on the subject. I was doing some tests with people showing part of my talks because there is some uh part of of screenshot from movies and to see if they were too old to be understood.
Speaker 2: As a disclaimer, I want to say that this movie I wanted to show you is part of my knowledge on the subject as a as a young boy and not supporting any any crazy idea connected to them because someone advised me against it In fact, the first movie I ever remember I ever watched connected to the concept of AI is this one I don't know if you ever saw it, maybe you know the name of the movie. It's the Stanley Kubrick movie. From 68 I wasn't I'm not so old that I was able to watch it on the theater, but I watched as a young boy
Speaker 2: and This is my first ever idea connected to artificial intelligence because I saw um a computer that was able to thinking on their own on his own and interact with people speaking and Also without any human presence. This is a sh uh a key scene on the movie. th you see in the spaceships the the the computer is uh j uh managing everything and uh This fact was a bit fascinating for me, but also a bit scary to see that a computer can behave
Speaker 2: on its own without I think a lot of people share this fear against AI or modern tools or new things. I collected some of them from internet and I show with show share with you some sentence I found. This seems a pretty uh strong one. I removed the the subject uh is an extract from a longer sentence but we can say that is a quite provocative it sounds very anti -AI it's quite strong Let's see another one. This is it's something that we can relate to, but it's also
Speaker 2: strong against code you can't create on yourself. But let's see who wrote this sentence. This is the first user. I don't know if you know him. is uh diestra is uh um the creator of the shorted path algorithm i studied in school and university maybe you know him but he's talking about cobalt In 1975. But it can be applied to to these days about AI. And this is the second user I I take a sentence from, maybe you recognize also him is
Speaker 2: Ken Townsom, is the co-creator of Unix and he was speaking about the code from its own company. in 1988. So as we can see things in our field are are circular. We see things happen more and more in in the same way. People are scared from new things. Maybe they are right sometime, other times they are not. There are people that jump on new fancy things and people that are more defensive. It's nothing new but every time there is something different that we we want to observe.
Speaker 2: But in the title of this the talk I spoke also about another part of the scene today, the maintainer. And when I think about maintainer This is the image every time I think it's quite uh uh perfect it summarizes everything is happening in our words. In this case you can remove the name of this city from uh this image and put your own city, a small city in Europe or elsewhere, and it works the same And we can also put there not only a small project, an open source project, we can also put
Speaker 2: our project that we like more, Django. I used this image since a long time. For the few people that know him, it's Django Reinhardt, a famous jet guitarist. and it's the one that the creator, the co-creator of Django Adrian, took the name from. It's a very great image. It's in public domain, so I put in every slide to talk about Django. But I found recently another image. More interesting. This one. Okay , not true. I
Speaker 2: haven't find it. I asked to AI to modify it for me. So instead of asking AI to modify Django, the framework, I asked AI to modify Django, the guitarist, and to see with you what bring to us. First things he put a very new modern guitar in instead of the Jets one from the last century. It's something like the ideas also in the code try every time to put modern things also in very old uh project like Django It can be something interesting, but sometimes it's not felt right. It does also another interesting thing.
Speaker 2: It removed the the cigarettes from from Django and also the smoke. This can be a good improvement, maybe for his healthier attitude. But the the more strange things he does it is to put m one finger more than the one that Django really had. I asked for the AI to improve the picture but it not in in in its own it decided that one finger more is more modern than the other one. And actually John Gorinard had some injuries from some uh file and it had
Speaker 2: uh very um uh i it wasn't able to move or the or the finger and uh in this way he created a very different way to play the guitar but for sure he had only five fingers. The last thing he added in the picture that we handed before is this one The amplifier is quite modern, I think from from the 90s or or less. So it's not so fitted in the picture. Maybe it's it look great there but it's not from the 40s of last century. But I think this is the very good uh metaphor or what AI is for me. I'm speaking about amplifier.
Speaker 2: More and more I read about it and more I think that the AI can be an amplifier of what we already have in our community. It can amplify the contributors we add because more people can write code easier, but it also can amplify the low value contribution. because more people can write this type of code. We can have more code, quicker, easier, but we can also have more untested code. So it amplify good things and bad things. I tried to put some my ideas in in in the slides. I am more and more convinced that AI don't change the nature of open source or our community, but it changed the scale of it.
Speaker 2: I'm saying this observing from this point of view, I've been in the community since a long time. At the beginning as an attendant like you, then I decided to be more involved. and being more different hats I'm seeing how AI is impacting different projects and uh different maintainer of the this project. The one that we care more is of course Django and especially for this just before uh my talk and the previous one uh another speaker uh Marcus spoke about the this topic more from the point of view of
Speaker 2: uh security. In this picture I'm JengoCon Europe 2017 during the sprints and Marcus was mentoring me in delivering my first uh big pull request in the Django code and he just s um uh presented a talk about the security in Django and says that the AI has an impact on the security team and can be uh something that we have to protect for, observe it, we 'll see. But other than this our community uh impact I I tried to
Speaker 2: explore more the metrics to find more more uh these cases of the impact of the AI on our community A lot of things are happening and working on this slide , reading news on Mastodon or other So shalize more and more of these episodes. The first one that I ever remember is from a project that I'm not personally use but I know is Okalm. And it's a very good project. ISO this this first pull request.
Speaker 2: with a very huge diff of code and the the thing I was observing it's was a pull request with a very huge scope on what you want to do or do. It was open and without any previous discussion and the review of it was of course impossible. What happened in this that pull request was that one of the maintainer was forced to to close the pull request and lock it because to avoid the going ahead of comments and debating on something that was draining the time in the of the maintainers
Speaker 2: We can show more I can show more and more of this one, but the next one I want to show you is something that scale from this This one. This is another project open source. I not use it, but is very well known. It's Godot, it's a gaming engine. And in this case I observed um that the pull requests became more than one. They tried to open A lot of pull requests without understanding exactly what they were doing or automatizing what they are trying to do. And the effort of reviewing all these pull requests impacted a lot with the maintainer and in fact one of the maintainers
Speaker 2: wrote a frustrated message on blue sky in this case I think without a lot of feedback from the community. So we moved from one pull request to a large number of pull requests in a project and it impacted more on the maintainers. The next project that was impacted by AI is one that maybe you know. It's the Jets Ben project. It's very overlap with the Django community and Janice the creator decided in this article to sunsetting hits In this case, the AI acted a lot as an amplifier because there are there already was some issues in the project
Speaker 2: that uh the the advent of AI accelerated or amplified. And in fact, in this long article that I only take one piece It said explicitly that these floods of pull requests confistening to sunsetting Jets band and starting to close. to close it so in this case it's accelerated what's was already going on something different this one maybe you know the actress But this case is the hardware of a contributor and a user on PyPy that created this package.
Speaker 2: I think in three days it's reached 30,000 stars on GitHub, something like that. But the point of this project is not clear if the creator is a real human. He hides behind the name of the face of this actress and uh with also a lot of new AI bots uh it's not clear anymore if a project is made by another human, the developer, or created from scratch by AI I think we need some something to recognize people and and droids. if you don't get the the situation is
Speaker 2: Blade Runner and they are trying long time ago 40 years ago to try to distinguish people and and androids This is another project that maybe you you know well, it's Matplotlib. And this pull request has been quite uh quite huge. An AI bot generated a pull request on the sub on the project. The maintainer rejected it for various motivation. It's quite similar than before, but in this case the it escalated outside the repository because the
Speaker 2: it moved to a conflict between the the i bot to the maintainer in fact The bot added a comment directly on the pull request, stating that he have written in his own blog an article against the maintainer of the project because it closed his pull request and the article is full of uh personal attack to the to the maintainer so it's scale out of the code and way away from only the codes. And this episode um was one uh
Speaker 2: thing that uh scared me a bit. It remembered me this scene from Space Odyssey. the original movie I I showed you when the machine for the first time refused to do what the human were asking for As we see, the problem is no longer the code or the bad code, it is the conflict that can be created between maintainers, developers, AI and And what I think we needs as a community to protect from this in some way And I want to show you some of this that already started happening in our community
Speaker 2: on in Connected Community I'm only showing you some extract from documents that have been created or rules that have been added to to some project. In this case is um are quite uh permissive rules that has been added in this project and they accept code from a generated from AI but they ask you to take your responsibility is quite vague but the complete document is more is more complete. This is from the Python new documentation. You can read the the world documents
Speaker 2: But more and more project as adding more restrictive documents in this case They discourage you to submit if you are going to open up a request it requires more effort to review than to write or to create. And this is the Fast API. project. This is a longer extract. Maybe some of you can recognize it They started before that unverified contribution is a burden for them maintainer and in some cases the submission without manual verification can be closed
Speaker 2: without review It's a bit stronger approach than the Fast API. This is the Django AISST contribution section in our latest documentation. So we are moving also in our community to do something about it. But there are communities that are doing something stronger. In this case, they stated that a low value contribution will be closed and because the more important thing is the maintainer time, not other things. This case is Matplotlib. This is the the same project that has been attacked directly from the AI that
Speaker 2: written an article against them. So maybe there are more. preoccupied from this type of episode. So the maintainer time spent to answer the comment is more valuable than what can be of value in this pull request. This is another um structure statement. The wall document is is more detailed, but they asked people to take responsibility for everything that they put in the code. This is another Python project arch that has been showed before by Cartoon and it's talk and if you go in the in the detail of the policy you can see that they request more strictier things than
Speaker 2: we are asking in the Django community This is the final level of the things. No AI contribution at all. This is forbid. This is from the Zigg project. Maybe some of you know him. So we are in a point when we have to decide it. What we want to choose, we can embrace the technology and try to manage it. We can refuse it. I don't think there is a uh I have a good answer to this. We are as a community we are seeing what other people are doing. We are discussing on it.
Speaker 2: It's time to to choose what what to do. It's not a simple decision but the the good things is that we started searching for it So summarizing up. As I said before, I think that AI is working as an amplifier of problems but also possibilities you can embrace it and do very great things but on the opposite people can embrace to stress the problem we have And so we have to scale our responsibility to share more this responsibility and to higher the level of the protection against this this problem. Since we are here in this this room in person
Speaker 2: speaking each other, I think that the key point is the human. We need more human in this loop to to share more ideas, something that the AI cannot do. And speaking Hoy Human, I want to Thanks all the organizers. This is the picture from the last DjangoCon Europe. I used this one because I don't have the final picture of this year. But I want to thank all the volunteers and the people that the people real people that are um organizing this thing and let us know be here Okay, you can download my slide later. You can also see the world links
Speaker 2: because they are connected to the world article and made your own ideas on the subject And as I said before, this is my contacts if you want to talk with me about the subject and share your comments or your criticism please not AI comments only personal thank you thank you very much Just before the question, I need to take a picture with all of you. with all these people in person.
Speaker 1: Can we turn on the lights on the audience?
Speaker 2: It can work. Okay. Three, two, one, go. Thank you.
Speaker 1: Thank you very much. Thank you very much, Paulo. Uh I was planning to put my assist and to do uh the introductions but then I saw your presentation so I said okay I'll do it in person. No thank you very much. So do we have uh do you have questions about uh the presentation? One on the back. Thank you. Not enough steps today.
Speaker 3: Uh yeah, thank you. First of all, I'm not an open source uh maintainer, so I can obviously not understand and feel uh your pain. Uh but just I would like to To react on the last slide when you wrote uh more human in the loop. I mean yeah of course that's ideal but uh I was thinking that to like to fight this this AI uh overflow in contribution maybe it would be helpful to use also AI to filter those those contributions. Uh I saw that there's a package that has been uh published some time ago uh to
Speaker 3: to uh detect uh if uh like uh pull requests on GitHub have been made by AI or not. Uh what is your opinion regarding this approach? Is it also something that you want to to use or that maybe you already use Um
Speaker 2: yeah, I I totally agree with you. What I'm trying to say in this slide was not to not use AI. I think using AI as a tool, as an amplifier or other possibility is is good We have to be able to use it properly. What I'm trying to say is that we cannot set up an AI bot and say fight this security issue that has been open or in their home. We need some expert people that know how to use AI and to build easier something that is more effective to to manage this new contribution, not to fight
Speaker 2: and block it, to try to distinguish. human that can use AI as a tool and not AI that act on his own. This is this is the point I think
Speaker 4: Thank you very much, Paolo.
Speaker 2: Oh hi. uh
Speaker 4: not to trust any code that you didn't write yourself, but um I don't trust the code that I write myself either. And so you know I have to start with protecting myself Against that. So first question is I wonder if there are some strategies that help save us from ourselves as well as from these other things that we're concerned. about. And secondly, while the AI slop type of contribution is a burden and so on for the maintainer Um I think that the AI augmented actions of bad actors could be far more serious. threats to open source software
Speaker 4: because now there is this inexhaustible amount of energy and and uh attacks that come from all directions that allow targeted attacks to come in more easily, deliberately, not just through laziness. But you ever thought about that?
Speaker 2: Yes. He co-created the Unix. So but it was not it is it also was afraid it was speaking about its own code that he wrote from his own company. So you share the same um preoccupation of Ken Thompson, so you are in a good good company. And speaking about the AI that can be more serious in the attack. And as I as I said, I personally use my daily job AI as a tool. Every time I need to uh fix something or be more precise in asking things and I don't
Speaker 2: have AI outside my things. I try to use whenever I can't I decided no not to use sometimes the point is as I said in the last the last slide is to having more and more human in the loop to be able to recognize this type of escalation and to decide it with our own approach. community by community to how to face it. Maybe speaking up with other friendly community open source and to find together something that we can uh use to not close uh only the the the door to AI contribution but to use it
Speaker 2: uh when we can and to contrast when we have to is something that is evolving so I don't have the the solution for every situation. I share with you different of this situation to say that it's evolving fastly, it's accelerating, is multiplying. So we want we need to speak about it and find something together.
Speaker 5: Hi, uh very thoughtful talk. Um at work, right you said when we write more with AI We build better static analysis, better linters and so on. We don't add more humans to the loop, right? We add even more machines to the loop How would that be for our open source contribution? Add more machines to checking before we even as humans look at the contributions
Speaker 2: Yeah, when I 'm thinking about maybe trying to explore, I'm thinking more about this one. This is I think the metaphor of a contribution to Django of AI. It's true that using this tool can remove the human from the picture, but actually if you think AI is a tool, It happened the same when we started using some ED instead of using Vim in the terminal or it 's another tool. We have to be able to use properly. and to know that if you are not very careful you can add one more finger your own project.
Speaker 2: You at the beginning you do not understand and the things that we have to be aware are subtle sometimes so the human in the loop can be recognized can recognize this if is an expert or if we have more and more eyes to look at that. And so we have to be there to monitor and see what we can take from it is valuable er or not
Speaker 1: one last fast one quick one
Speaker 6: Thank you very much. Uh very quick one. Using your uh metaphor with Morpheus, giving two peels as a choice. On the one hand, we have at the very moment the maintainers open source code maintainers uh being quite strictly against uh the AI generated quota. On the one hand On the bad one, uh in my personal opinion, we have the big tech pros uh stupulating, like for example the the Shopify or Spotify was it, um who uh told us that like year
Speaker 6: last year they didn't write uh they didn't write any lines of code themselves so they used only the AI generated code In your opinion, who's gonna win then? Um the open contenters will like just accept the LLM generated code or will the big tech bros come to the good side?
Speaker 2: I think from my experience, from what I studied the history already, Answer has this question. I remember reading some tech people in the seventies saying that obbyests cannot be able to produce any valuable code was Big is speaking about uh hobbyists writing and hacking around or speaking about Linux or open source and we saw that history proved it wrong So we are here because we share and care about open source software that is also an open community and I'm not saying that we don't need
Speaker 2: companies that do business with tech technology, but on the opposite there is no one not one only winner on it, but being and involved in a community and being part of a open source project is in itself a win. It's something that is not possible to think about. 20 years ago and now there are plenty of open source community, plenty of people that write codes, very valuable codes And other projects like Wikipedia and people were saying that it was untrustable. Now they are writing their encyclopedia with AI and Wikipedia is better than this type of of data.
Speaker 2: So I think we still have um the opportunity to be valuable. As I said at the beginning, AI don't change the nature of open source It only scale and multiply and problems and maybe advantages and It's something we want to see in a few time, but I think if we work together and start thinking about these topics will help us to go ahead
Speaker 1: Thank you very much, Paolo. I think uh we're out of time, but uh feel free to use the chat if you want and keep discussing. I'm I'm sure that Many of us have many opinions to to share and discuss about this topic.
Speaker 2: Only one request, the last one. If you want to send me please a Oh okay. Send me please your comments, your requests, your picture if you take some of the the talk and I'm here around uh to talk with you all. Thank you.
Speaker 1: Thanks again
April 15 - 17, 2026
Athens, Greece
Note: We understand that names change, people change, and bodies change. We respect each individual's journey and privacy. If you have any concerns about a video or need us to remove content, please don't hesitate to contact us. We will handle your request with care and promptly address any issues.
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026