Proposal for an AI team for Django

[thibaudcolas]

This is a draft charter for a new Artificial Intelligence (AI) team for Django, acting as an "interest group" - a forum to exchange ideas and coordinate interested contributors. The goal of the team is to encourage thoughtful adoption of Artificial Intelligence by Django users and developers.

Why

We need to guide our community towards better outcomes with AI. The industry as a whole is busy with adoption left and right, we need to understand what it means for Django users. Fight or avoid the problematic aspects of AI across our industry, and showcase the "good parts".

How you can help

It’s a vast topic, we need feedback on this charter, suggestions on possible activities for the team, and more members :) As this is a popular topic we will try to make this work with a large number of people, TBC whether everyone is a member of the team or if we have small group of team members coordinating contributions amongst a bigger group of interested people.

If you want to join, please take a moment to:

1. Review the proposed charter in this PR and share any feedback you have.
2. Submit our Apply to join the AI team form.

Timeline from my side:

1. Get reviews from interested members to make sure the charter is in a good enough place
2. Get input from the Steering Council (@django/steering-council could you take a look at what is proposed here and share your thoughts?)
3. Get this to review by the DSF Board

---

Thank you @jangondol @sheenarbw for writing the first draft of the charter with me at the DjangoCon Africa 2025 sprints!

[czue]

I'd be interested in being a part of this!

[wsvincent]

Also interested. My talk at DjangoCon US is all about this topic and ways Django can integrate AI more.

[nanorepublica]

I'm interested, but perhaps just as an observer right now (I have too many other plates spinning in my life!)

[SomeAB]

I'm interested as well, I recently floated an idea related to ORM optimizations that could work well with AI

[ulgens]

I recently floated an idea related to ORM optimizations that could work well with AI

I was playing with similar ideas myself. LLM / AI-focused operations, such as tokenization and similarity search, can be implemented at the model layer in a way that allows third parties to be used.

That being said, I'm more interested in the ethical usage of AI tools and not abusing other people's data and natural resources.

Count me interested 🌞

[wsvincent]

As a next step, perhaps those of us at DjangoCon US in a few weeks can meet and discuss?

[blighj]

I'm interested.

[i-salameh95]

I’m interested in joining! This would let me apply my Master’s studies in AI and also contribute from my experience as a Django full-stack developer.

[awais786]

I am also interested.

[NixonNelson22]

I am interested to be a part of this.

[thibaudcolas]

Such a great response, thanks all ☺️ if you’re interested, please:

1. Review the proposed charter in this PR and share any feedback you have (or "Approve" if it’s all spot on)
2. Submit our Apply to join the AI team form.

As @wsvincent suggested, for those who are at DjangoCon US 2025, let’s meet up there and discuss this!

---

Timeline from my side:

1. Get reviews from interested members to make sure the charter is in a good enough place
2. Get input from the Steering Council (@django/steering-council could you take a look at what is proposed here and share your thoughts?)
3. Get this to review by the DSF Board

[nanorepublica]

Could this team become a point of expertise within the community for other Teams/WG's to contact for advice?

[thibaudcolas]

I think yes? Do you think this should be written as an explicit responsibility?

[nanorepublica]

I would say yes, so that other there is an expectation and possibly some coordination between teams. While this likely happen naturally, to me we would benefit from more being written explicitly rather than assumed.

[czue]

Just a few small suggestions

[colleendunlap1]

Interested in joining!

[thibaudcolas]

@colleendunlap1 excellent :) I’ve just set up a new form: Apply to join the AI team. Please give it a go! @joshuadavidthomas thank you for being the first to test the form 😅 I’ve now fixed this hopefully.

---

For everyone else – I’ve rephrased key aspects of the "Scope of responsibilities" based on feedback I received, to be clearer the main goal is for this team to coordinate not just people who are enthusiastic about AI but also those who are concerned about its fundamental issues. There’s lots of aspects of AI adoption / Django suitability for ML/LLM apps that could happen with or without this kind of coordination team, but for some fundamental aspects of the project it’s essential we can have tough discussions. We really need people who think AI is a net negative for society to get involved.

And if anyone has concerns about this team but doesn’t feel comfortable sharing them publicly here, please email or DM me, contact details on my GitHub profile.

[trey]

Count me as one of the folks concerned about "AI"'s fundamental issues and who thinks of it as a net negative for society.

I'm not surprised to hear the typical AI hype in other communities, especially those that still frequent Twitter, but I'm disheartened to see it getting louder in the Django community.

I don't know how much my voice counts in all this. I've never contributed to Django, but I have been using and loving it for most of its existence.

If you want me as a part of this as a dissenting voice, I'm glad to weigh in for a little while. I don't intend to be the one voice of reason, though, so I hope others like me raise their hands as well.

[thibaudcolas]

@trey yes please 🤘 we definitely need dissenting voices / AI critics involved with this (not sure if that’s the right term?). To challenge some of the enthusiasm. And to help us with the harmful aspects that lots of people are unaware of, or ignore.

Please apply to join the AI team and if you feel like it review the proposed scope here or send me feedback in private via email or Mastodon (see profile).

[…] I'm disheartened to see it getting louder in the Django community.

You’re not alone. There are lots of aspects of the current generation of AI tools and adoption trends that are at odds with the Django community’s values. I think it’s a good time to get involved.

---

Linking to recent / current / upcoming work in this space for people who are interested in this:

• WIP PR to add AI usage disclosure to Django PRs
• The DjangoCon US 2025 programme had excellent talks about "AI and Django". Both enthusiastic and dissenting
• For the upcoming DSF Board elections we’re working on we want to discourage candidates from using AI.
• #ai channel on the Django Discord server (current topic: energy use)

[tim-schilling]

I have concerns around the phrase "Making it easier to build AI/ML products with Django". Using this phrase in the charter of a team implies that changes to Django are within the scope of the team.

[czue]

I wonder if it could be rephrased to remove that implication? E.g. I think the AI team could serve this goal via documentation, tutorials, etc. without having to make any changes to DJango.

[tim-schilling]

I'd like to see this group be re-imagined as a time-boxed research / discussion effort. I think there's benefit in reviewing what parts of AI/LLMs, align with our ethos, if any and determining a path forward.

I think there are a few discussions that could move forward simultaneously:

• To what extent should AI/LLM tools and information should be included on official channels
• What is the impact of LLM generated code, tickets, security reports on the Django teams
• Where can tooling be built to produce a better developer experience for people building with Django
• Where can documentation and training materials be created to improve the Django learning experience

[jacobian]

To echo and agree with @tim-schilling - I also think this needs to scoped carefully. The potential scope here is massive, and the ground is shifting rapidly. There's a very real risk that an AI Working Group could turn into a morass -- a never-ending sprawling conversation, or. worse, a series of heated arguments. I'd like not to have that happen!

My suggestion:

• Explicitly scope this as a research and scoping project for future work. No building, no tooling, no new policies, etc -- purely an effort to determine what the DSF could and should do. A major component of this should be talking to the community, gathering input, and summarizing the results.
• Explicitly time-box, as @tim-schilling suggests.