We've been getting a lot of AI coded submissions recently, as well as lots of reports asking to remove them.

Should we remove these posts, or not?

Rule 1 says Posts must be relevant to the open source ideology

I don't think any ai projects can claim to be that when their copyright status is debatable and the tooling is worse with how closed their training is.

Definitely need to remove them.

I vote remove

Remove, and suggest they go start an AI software community. We can't be monsters to people who might not know better.

Remove them, let them create their own "AISoftware" community.

I think a disclosure at least in the title of the post should be required. But only if the software is known to be not slop at all. The problem with this approach is, that everyone has their own threshold and definition of what slop means, or are misinformed about the project. Overall I do not think it's a good idea to ban any Ai contributed or assisted project.

As an example, do you consider Linux to be Ai slop project?

Remove them!

Remove them. There could be another community for slop.

AI slop & AI code is very different. If its just pure vibe coded slop with no oversight or effort, then sure its just junk, remove it.

But a lot of experienced developers now use AI tooling to increase productivity, just because AI was used in the process should be consider it slop?

I would say to just require them to be tagged, but the volume of them seems quite high. Might be better to just ban them and they can create a separate community.

So no Linux news anymore. :-(

I would say if the post itself is entirely AI generated I'm not interested. If it's a human post talking about a project created with AI I don't mind so much.

I think requiring AI disclosure like how !selfhosted@lemmy.world does is a good approach. Not all projects rely as heavily on LLMs as others.

Disclosure. If LLMs are used in service of open-source projects, then they are relevant, but clearly many in the community feel that to be negative, and therefore there should be a disclosure. Someone can make open source software using closed-source tooling already, so disallowing one form of tooling entirely is letting a moral reaction interfere with how we understand open source.

Linux would have to be removed from discussion, for example, if all LLM-contributed projects were to be disallowed. I don't see why rules should be made with explicit exceptions in mind, yet excluding Linux cuts out a huge portion of open source discussion.

Keep them but perhaps ask for #ai tags in titel, so those not wanting them can scroll past them.

All these new apps/software announcements will just get better and better, and if they are open source, they clearly fall in under this sub. At some point most users will have their own agents that can add or improve these new projects as they see fit, so let si oss be shown here as normal.

There are already super cool vibe coded open source software, and let's be honest, not all hooman software is great, so why not ban shitty hooman software/coders while we are at it? Oc not - oss is oss no matter how its made, even if the description sounds like an ai, or it's not always maintained.

In the near future, pure hooman coded open source will become more and more rare, so those interested in non-ai open source should start new sub's for their special wants instead of posing restrictions on what is becoming the new normal.

Remove. Perhaps there's another community interested in AI hobby projects, but most of these seem like listening to someone's dreams, and are not very interesting.

You are just adding your oss preferences to open source here. I too have preferences, like no corp oss, no ads, no logins, no telemetry and so on, but i have to accept oss as oss more broadly.

I would recommend that you set your agent to filter out posts you don't want, but you would likely rather censor this sub than automate it with an agent..

Not calling it slop might help get less biased answers.

I'd like to keep them, but properly identified as such. As much as I don't prefer it, without it there would be projects that wouldn't be attempted.

They are also used by proper developers that know what they do and they just need specific parts done, which they'll review anyway.

So... Keep them. Identified, but keep them.

From my perspective "slopcode/vibecode" means entirely (or mostly) generated, which is different than a human-maintained project that includes bits of generated code. As a rule I'm very skeptical of the former and give the benefit of the doubt to the latter especially if it's an established project.

I'd be fine with removal or mandatory disclosure, but the latter would probably be easier to implement if there was a way to filter them out.

You are just adding your oss preferences to open source here.

OSS == Open-Source Software, what are you talking about

I'm mostly referring to these new projects that are AI coded from scratch, not existing projects that might have some LLM contributions.

In my view, starting code from scratch or augmenting existing code with LLMs are equivalent in terms of whether or not they should be eligible for discussion. Both should have a disclosure, and both are worthy of discussing where relevant. If someone uses a bot to advertise a million LLM-started projects or something and overwhelm the community, then this is bad and should be stopped, I suppose.

I'm referring to these new projects coded by AI from scratch, not existing projects which have an occasional LLM contribution.

People are submitting a lot of these AI projects recently.

Agreed. Needs to be disclosed in the title of the post as having AI involvement, and show signs of active maintenance. Anyone can now oneshot some slop that's nothing but a waste of our time, but it takes good developers to maintain, secure and expand.

5 days old, nope, hasn't been touched in months, nope. Dozens of active human participants, and/or a reasonable history of active development and to a lesser degree stars or equivalent, sure.

It's a bit more work for a poster, but I think it can keep quality high for project posts. Should be voted on by the community and made a rule.

I would recommend that you set your agent to filter out posts you don't want

Just to be clear, do you mean that the user's user agent (lemmy client) should be capable of filtering out posts about vibecoded slop, perhaps by having lemmy communities require such posts to be tagged? Or do you mean that a LLM should curate what users see here?

The question is how opensource is AI model which has been used for these creations? Are datasets available somewhere?

Totally agree. The project has to be developed by someone that knows what they're doing.

Remove them. If you can't ensure you own all the code written you can't make it open source. It's not yours to license.

I hope it's not allowed here

Removed. AI is fundamentally anti open-source.

I advocate for a disclosure, blanket removal feels like censorship; even vibe coded slop is a project that might generate interest if it was made to solve some problem or showcase a new frontier, nature of OSS will push the slop out of the project eventually.

When I say "slop", I refer to a huge codebase that does what it advertise but has no documentation on what it does internally and the primary poster cannot adequately explain what it does(code mudball esentially). but if its well documented and looks like it was put together with care, with rhyme and reason than I will defend against calling it slop

I don't think restricting what people can post is a good idea if you are looking to grow a community

You're missing the point he's trying to make IMO. He's questioning if software can be categorized as open source, if it is coded by closed source LLMs, that in turn only aquired their database by stealing someone elses intellectual property. It's a valid question. OSS is more than just "I can see and fork the code".

Posts must be relevant to the open source ideology

Did you not read the full rule?

How can something that is inherently closed sourced be in line with that rule?

How about making a rule that it is at least six months old and in continued development? This would filter out most slop.

People against LLM parrot the same arguments again and again that it has simply lost its original meaning.

copyright status is debatable

debatable by who? who is going to settle this debate?, the courts who serve corporate greed?, Legislation made by politicians lobbied by SaaS companies? Anthropic? OpenAi? Microsoft?

The "Copyright is debatable/ its just a plagiarism machine" stance is not the Leftist/ pro FOSS gotcha argument you think it is. The point of the FOSS movement was to have total freedom over software, and that freedom includes the ability to study, learn and use it in ways it was not intended for. To stand as the antitheses of that is make Steve Balmer happy

What about project that were developed normally, but later changed to a complete vibe coded slop model? Because if this is allowed, then this would be a loophole basically. Just start a project without Ai, then do all the stuff with it.

Tell that to all the proprietary software that got OpenSourced

We had Slop before AI. we just called "shit".

If a coder wants to shit out a project that never works and never patches it? I want them to notice the culture and know better.

If someone is learning, using the tools they have available, and shows a commitment to maintain their project? I want to see their work. I wanna see them get help from the community to be more literate in their own project, and in the full toolset they're using.

Most of these vibecoded submissions are the former. Thank you for removing the fecalware. If the community decides it needs to remove AI at all to keep this up? Eh, there's dbzer0.

Oh you mean the broken "replacement" for Photoshop?

The one that wasn't needed since we have GIMP and Krita?

Yeah that was fucking worth annihilating whatever teeth the GPL and associated copy-left licenses had left.

I've recently been pondering the future impact of AI on coding, while trying to filter out the current noise surrounding the issue (since all it does is muddy the waters). There's a lot of amateurish supposition to follow, but since you appear to be getting fewer thought out responses than "AI-bad-kill", take it with a pinch of salt.

There was a recent article on the open source software that props up the internet - (paraphrasing here) one of these repositories had (pre-AI) a user called Jin Yue or Jin Yin who made hundreds of useful pull requests, gained the trust of the repo maintainer and was eventually given admin privileges, only to surrepticiously merge malicious code when no one was looking (eventually discovered ~a month after the fact). I imagine that (likely) state-sponsored attempts like these will make more use of AI in the future for quicker turnaround, and that maintainers will also eventually use AI to quickly audit submitted code for security flaws.

As far as open source software channels are concerned, I imagine there will be a decent amount of low effort malicious software that makes heavy use of AI, submitted to these channels. In such an environment, whether or not users are permitted to candidly disclose AI usage without reprisal will likely make a huge difference, since honest disclosure would invite security-minded people to audit fault-prone AI-assisted code - something that malware distributers will try to avoid at all costs (thereby allowing a distinction that can be leveraged to form between the two).

Just my two cents.

its a movement, i dont endorse the projects themselves but it sends a clear message

I’ll tug on that thread… so can we not use Windows or macOS to develop OSS code on? It has to be on a Linux machine? What if the BIOS is closed source or it’s using binary blobs for the video drivers?

I think it should allow AI coded projects.

AI code is a QA problem.

I've seen an opensource project (amule) being revived with AI, but there's extensive testing before the PRs are merged and the project seems to be working pretty fine.

Btw, I use AI a lot to configure my emacs init.el file :-D

We already sent it, and it was with Krita and GIMP.

Fuck the message "we don't need developers", the message was supposed to be "programmers don't need corporations."

I am going to go the other way and say the more, the merrier. I like to see new and interesting things that someone did to scratch an itch or solve a problem. I may not use it or agree with how they developed it or the license they chose, etc., but I still like seeing the posts about it.

Very hard question. I think, idk. For me Open Source isn't AI so i think remove it because it makes /c/opensource less attractive and worse but some things like about Craft i think it can be but i'am not moderator so i think ban it completely or most of it.

Since when has development tooling mattered for something to be FOSS? If it relies on propietary software to build or run it then there's an argument it's not truly free.

But if we're saying that anything developed using closed source tools is not allowed then we'll have to throw out anything that was developed on a mac or Windows PC.

I'm gonna agree with others that there need to be standards not a blanket ban. I'd leave it up to the admins/mods what those are. Because getting too granular with it wold be hard to enforce. You guys pick something you feel is reasonable and will cut down on 0 effort slop but also allows you to enforce things without digging through the minutia of every post.