Community-led open source project hosting site Codeberg has formally announced that projects whose code is largely or fully machine-generated through LLMs and other ‘AI’ tools will no longer be welcome. This follows on the heels of a similar ban on cryptocurrency-related projects.
The community vote was on two issues, the first being the notion that scraping of project code for the use in LLMs should be forbidden, which was a motion that easily passed. The second motion was on disallowing projects whose code was substantially generated by LLMs like Claude, OpenAI Codex, and similar. This motion passed with 358 in favor versus 144 against.
In the earlier linked blog post the reasoning behind especially this second issue is expanded upon, covering not only ‘license whitewashing’, but also the direct and indirect hardware costs, with the expanding ‘AI’ datacenter hyperscaling having massively increased hardware costs for Codeberg over the past years, as the costs have been largely externalized.
Also covered is the aspect of these LLM-based tools destroying the OSS community, which is something that is backed up by recent studies. Even if we ignore that such LLM-tools are destroying the cognitive abilities of its users, there’s an argument to be made that if LLM-scraping is disallowed, then it’s consistent to also not allow LLM-generated code.
In the Terms of Use you can see these changes, both for LLMs and for cryptocurrency projects.
Thanks to [mk-fg] for the tip.
This anti-AI critique is following the exact same pattern as what they complain about: a torrent of low-effort, shallow contribution. – If anyone actually brings up a solid argument against AI, at this rate it will get buried in the sub-critical noise.
I would not be pro-AI if the anti-AI front actually had its act together. Stop flooding my airwaves with noise and we can actually have a two-way conversation about pros and cons and nuances and tradeoffs. Until the signal clear up I’m not going to even listen.
I fear I’m being pushed in the same direction. Nothing has pushed me from liberal back to more moderate than watching the same mindless, reactionary, uninformed drivel and outright lies about AI. I’ve come to expect from strong conservatives, but it’s really showing up on both sides of this topic. I find it especially ironic when people are doing that on Facebook or similar, where they’re literally feeding the beast they say they hate.
Making measured, informed decisions around technology is important. The lack of honest discourse on this topic is both frustrating and leaves me feeling kinda gross. In the meantime, I have no option but to make decisions based on what I know about the industry and my professional expertise.
The main arguments against “AI” generated code in open source is the possible pollution from projects with incompatible licences and the fact that code (or anything) generated by a non-human cannot be copyrighted and that could call into question the licence of any project with a decent amount of machine generated code. What is wrong with those two issues?
Huh? What? Cause humans never do this? And it’s not like there are ways to scan codebases or anything lol.
This is utter tripe. I can take a picture with a camera and I immediately have copyright. You gonna tell me that pushing the little button was more “human generated” than typing in some prompts?
1) whataboutism 2) copyright is a scam anyways
2 is not true, at least not in the European sense. Copyright requires “artistic effort”. People who photographed old (public domain) paintings, then claimed copyright. The court decided that those photographs required conciderable technical effort to eliminate any artistic effort.
If you take a photo of someone’s code, the OCR that, you have not created code that is copyrighted by you.
Also, “works of art” can have multiple copyrights. If I copy your code then modify it, we both have copyright and then we both have a say on how to be able to share it.
That’s a whole “derivative work” thing, and it perfectly illustrates how copyright is a scam. Is writing code “technical effort” or “artistic effort”? The court cases in question concerned conflicts of copyright – it takes just as much artistic effort for me to take a picture of my floor as it does to take a picture of a print of the mona lisa, but I will be recognized to have copyright of the picture of my floor! Same technical effort, same artistic effort.
But, this is also a side-quest – the original comment was about “non-human generation” not being copyrightable. Whether I’m pushing a button, prompting an llm, or writing a script to generate fractals, I can copyright all those things.
If you spend 5 seconds drawing a goofy original character on a napkin, that drawing is protected immediately. Copyright requires (some form of) originality, a (minimal) amount of creativity and the work has to be fixated to a medium in some form.
Just barely looking at the media industries and software industries will show that copyright, as it is currently implemented, is a scam built to let the rich hoard our culture and step on the little guy.
Humans are legal persons, you’re better chatbot is not.
That’s precisely the point. The human running the chatbot is responsible, not the chatbot. So if humans do the same things with or without chatbots, the complaint is spurious.
Open source licenses depend upon the enforceability of copyright. Open source projects depend on open source licenses. AI has the potential to break the entire paradigm.
“Huh? What? Cause humans never do this?”
They do.byt they take responsibility for it. Who takes responsibility when it’s generated by an LLM is not yet tested.
“This is utter tripe. I can take a picture with a camera and I immediately have copyright. You gonna tell me that pushing the little button was more “human generated” than typing in some prompts?”
According to courts it isn’t tripe. It’s tested law. A photographer frames a shot etc. they’re the “creative.” Writing a prompt is just making a commission. The “creative” is the machine. Anything made by a non human isn’t copyrightable. Like paintings by elephants or photos by chimps.
“1) whataboutism”
You don’t know what that means.
“2) copyright is a scam anyways”
You may believe that. I doubt you would if someone pirated your proprietary code etc but that doesn’t matter. What matters is we have to live within that system unless it changes and have to try protect ourselves from its currently rightful applications and the misapplications oh too commonly used by those with deeper pockets.
Non-issue, just push through it
we appear to be throwing our whole economy away on a bet that LLMs can replace most white collar workers. but LLMs aren’t smart enough and white collar workers need to check the work of LLMs or else a bunch of tiny bombs are placed in our IT infrastructure. to the extent humans don’t (they were laid off or are just human-levels of lazy) that seems like a huge and very widespread mistake. and now our pensions are invested in this terrible bet, the countryside is covered with loud ugly buildings, and nobody can afford to build semiconductor-based devices because most of the output of our semiconductor industry is directed towards this madness. the people who did this have a lot to answer for. but okay, tell us how LLMs are going to make the world awesome.
Do you live in Monaco or something? You should get outside more if you think data centers are an appreciable fraction of the landscape.
Hey maybe fix that problem instead of whinging about LLMs
So what you’re saying is, when the investors’ bet doesn’t pay off, we will suddenly have much cheaper semiconductor devices due to excess capacity. You can’t have it both ways XD Either they fail in the long run, and we benefit from the glut, or they don’t fail in the long run and were right.
Taxpayers of a region will also be stuck with funding powerplants that were built specifically to supply the datacenter that went bust…..and no thats not cheap.
Again, fix that problem! Why are tax payers footing the bill for a powerplant for a datacenter?
This is a comment section and not a discussion platform. Better move that discussion to Lemmy or something.
You’re gonna need a bigger boat.
yet, here you are?!?
You don’t have to be pro-AI if you’re not anti-AI. Also, AI is fine as long as you write the code yourself.
another thing is: you can use LLMs (calling it AI is a stretch) without supporting the huge economic miscalculation being made by capital and governments who think that LLMs will replace the workforce and thus be a capitalist dream come true and will do literally anything it takes to bring that about
This particular argument seems pretty darned logical and entirely complete for the group in question, and really applies to most everyone that isn’t a really really giant tech company – “”AI” is eating up all the hardware resources, and we need some too which can only be had for ruinously inflated prices, while hammering our servers looking for data to ‘steal’ pushing up our costs again.”
Very valid points all though the document, which then leaves you with two real options in this case.
– actively enable more use of the very same tool that is killing you.
– do what you can to push back by not actively enabling the thing that is killing you on your own platform…
Really not much choice for this platform when you phrase it like that is there! As while yes these “AI” tools can do some impressive things if used with the right care, that doesn’t magically wash away the insane energy, water and hardware demands, that ends up putting up everyone else’s bills on these things too!
“company – “”AI” is eating up all the hardware resources, and we need some too which can only be had for ruinously inflated prices”
5 years from now when there is a more accurate portrayal of what’s really going on with manufacturing of components this argument is going to be seen as fundamentally wrong.
What we really have going on right now is a small bubble (contrary to the hype train view of it) in resource allocation, and an over inflated view of its impact being perpetuated by companies that stand to make profit off it. There’s been strong hints that manufacturing lines are intentionally reduced to inflate prices. Again. (This has been done before.)
Right now it’s just fits and sprouts of info that seems odd, but when the rest of the details come out in a couple years I expect we’ll look back on this, point fingers at a couple of key players taking advantage of it, and then they’ll pay a small fraction of their excess profits in a fine and carry on business as usual.
AI is next generation of RAD. It’s here to stay, and, done right, it is a boon to development.
Given the race to build data centres for ‘AI’ right now is costing more than an inflation adjusted Apollo program by miles, and that is just for the openly published information. With every time you look even more money thrown on the pile to crazier and crazier degrees, while so far many of the “AI” companies haven’t even turned a tiny profit! Then for good measure all this increase in spending isn’t reducing the ever increasing power demand either! If it was costing ever more but the efficiency was going up to keep resource consumption more rational that would still look really really bubble like, but at least be less annoying to everyone else.
Just because its a bubble doesn’t mean the technology is flawed – as seen with Railwaymainia or the dotcom bubble companies collapse the economy as a whole, especially this time with how integrated the “AI” have become to so much of it will wobble, but in the end the technologies survive and stay useful.
But really none of the end results of this is relevant to anybody providing a service that is seeing their servers hammered, no money coming in from all this work, and are unable to
get new hardware – for those folks even if “AI” is going to be a benevolent god level of useful to humanity practically tomorrow looking after themselves while waiting for that day requires some resistance.
I’ll give you a hint… if you approved every single Datacenter currently looking for approval… how many do you think would actually break ground?
Hint… it ain’t even 50%
And there are quotes from folks like those at M$ that say they are sitting on hardware they can’t yet use for lack of power/building to put them in… Even if they don’t build every single datacentre they are trying to on paper it is still throwing insane amounts of money and hardware at this pretty much non problem.
The argument against AI is clear: AI slop. That’s why the ban is against “substantially” or “fully” AI generated projects, an not “AI-assisted” ones–they’re targeting AI slop. It’s clear to see the signal, so why do you see noise?
Good to hear it. Codeberg stands for free software, and the rights of free software developers, and the freedoms of free software users, are directly undermined by how current generation LLMs are trained and operated.
I am not against LLMs in principle, but the current AI industry norms are unethical and bad for consumers – undermining software rights, raising hardware prices, pushing water pollution and grid strain onto communities.
They are going to infringe on licensed code about the same amount as a human would. Also, this whole “AI can’t invent new things or make new art” is completely predicated on the belief that humans create new things entirely without influence. Ask a person born blind to draw a sunset. Ask a person born deaf to hum a song. I, like all of you, was influenced by my environment and the natural occurrences in it. My “style” of art is not unique, it is created from the styles I have been exposed to. Just like learning to read one passage will not make me an author, but reading hundreds of odysseys suddenly will—and I can only write about what I know, as I can not already know what I have never discovered. AI trains from this the way we do, it’s just more specific in its execution and people can’t stand it.
Listen to this Ai generated music: https://post-man-pat.bandcamp.com/track/butterfly-ghost-dream and tell us if you think Ai is creative or not.
AI clearly creates things, it’s just an argument over whether any of it is any good (some is good, most is trash, but this Pareto distribution in art is as old as humanity)
…or it’s people uselessly splitting hairs and gazing navels about whether it’s “truly” creating or innovating, or if it doesn’t count count for philosophical reasons (meaningless, boring)
Yep, it still needs a human with ‘good taste’ to guide it.
Again, a LLM is not a legal person. Which is exactly used by thre LLM-providers as a excuse to infringe on licenses.
Humans cannot create new things entirely from nothing, but they still add their own personal perspective and unique connections.
LLMs cannot invent anything new because they’re built only to imitate the training data. Anything new it generates is by adding random noise, without critical thinking and most likely resulting in error and hallucinations. This does not model human invention or innovation, it’s just a random game of “Mad Libs”.
And how will you judge the signal quality with your fingers in your ears?
By protecting our hearing and not going deaf from the noise
With my handy-dandy signal-noisometer that I built
Pet-peeve: let’s stop calling any of this “AI”. There’s no I in any of this technology and “real” or “hard” AI is always 30 years away. It was so 30 years ago.
That being said, non-generative reasoning LLMs are an amazing tool for so many technical and scientific folks, and have been for some time. You’d be crazy to take these tools from mathies, physicists, and scientists. And than includes computer scientists. There are costs, of course. These tools need careful stewardship, and they are, all things considered, very inefficient.
But that cat is well and truly out of the bag; they have been in use for many years now — over a decade in some cases — and that will never change. I don’t think it should, either.
Generative AI or the trivial use of these reasoning LLM tools for “vibe coding” is very little bang for our buck. And given the great majority of projects on these repo sites are (sorry, not sorry) mostly unusable and outdated junk, a tool that serves to only increase those levels of junk is not welcome. So I see the point of banning it.
But I will continue to use these tools as a professional, treating it like a sort of pair programming. My job as a developer isn’t necessarily to personally churn out lines of code, but to solve problems and ship product. So I’m not making the argument I’m doing cutting edge research. But I am orchestrating a solution, and having an agent churn out the scaffolding and find some of the edge conditions while I do the part that humans are better at is just a win. The fact that I get a nice markdown file at the end of it that outlines the risks, effort, and cost for others to see what has been done is actually better than any human could do.
FFS, let computers do what they are best at,
Of course, there is the greater cost for us all, and we can’t ignore that. But this has ever been the way for the last N decades. Maybe not everyone should have access to cars, air travel, cloud computing, LLMs, the fastest silicon, quantum devices, and whatever is coming next. But this is a human problem, not a technical one. These things usually find their limits, though that process can be non-intuitive and lumpy.
Painting all of this with a single brush does feel like pointless noise, as @ganzuul says. We have to have honest conversations about this stuff rather than share an easy and obvious opinion. We are human for Crom’s sake, and that means we are supposed to be blessed with nuance.
So, it;s ok to say what they are doing here is fine. It;s their site, and the benefits of reasoning LLMs are not obvious in this context. But it’s also ok to think about there these tools are a win, and have that conversation.
Let the computers make the easy decisions. We need to save the hard questions for us.
I think I can mostly agree with this. Anyone who completely writes off the usefulness of LLMs after one has been used to generate a new mathematical proof on it’s own, is deluding themselves.
I recently used an LLM to generate code in an area that I have no experience in. I went through all the code and got understanding of what it was doing, cleaned it up but it mostly got everything I needed done and I learned some new things.
I absolutely wouldn’t suggest relying on it with no guards, but the fact that it could do chunks of code for me that I then looked through and it all looks correct and testing shows it just works, means that it has to have some level of usefulness.
I think you’re mistaken. It’s not LLM eho are used for mathemstical proofs but other forms of learning or beuromorphic techniues (also in the “AI” group). LLM have more limited use.
Aia, sorry for the typos.
You are just wrong about this.
https://www.youtube.com/watch?v=ZJ8KThKAfbs
The scientific stuff that has been used for a long time isn’t LLMs. They’re both applications of ML but not the same. In fact the optimisation of hardware purely for LLMs is starting to hold back the scientific uses of ML.
That might have been correct 6 months ago but things are changing fast. Agentic LLM with a harness and tools plus a virtual environment to run scripts and code it’s written, is now a valid approach to finding formal proofs.
Terrence Tao just proved the Jacobian Conjecture through a long ChatGPT web chat! However a more practical method for most people would be to run it in a coding harness.
Not at all belittling that but rhat’s one successful application of an LLM for an abstract pure mathematical proof. Not the raw number crunching needed for a lot of science and engineering. We don’t claim analogue computers are better everywhere just because there are niches they rule supreme. And one successful use (if it pans out) doesn’t even show that yet.
The hardware has also pretty stayed static or gone backwards on high precision workloads the last couple of generations because of the focus on low precision workload’s performance. I am not anti LLM (well except for some academic reasons to believe they are not the best option and smaller models will win out in the long run) btw. I am anti some of the industry’s business practices and putting all our eggs in one basket.
You have an exactly 0% chance of ever influencing that lexical juggernaut, sorry but your pet will continue to be peeved.
AI is always AI until it is well-developed and complete, then people simply call it “software.”
“There’s no I in any of this technology and “real” or “hard” AI is always 30 years away. It was so 30 years ago.” Current estimates of the next generation of Ai is about 7 years. Not only will it have logical intelligence like we’ve got now, but next level ‘real world’ intelligence as seen in animals like the lowly fruit fly. It’s already in existence in Intel and IBM research labs. Hot tip: Buy some Intel shares now whilst they’re still cheap.
lol Intel, the folks who brought us the Pentium 4! i think Intel’s only real innovation was clawing back the rights to the 4004 from Busicom.
That’s why they’re cheap !!!!!
I’m a bit torn on this. I can kinda understand banning wholly or primarily LLM created code bases.
However, there are some people that I really respect in the tech sector that have made good use of LLMs and support their responsible use.
So I’m always a little wary of attempts to stifle it because it’s very easy to get caught up in the anti-LLM zeitgeist and over restrict things.
They’re holding the lost position. It’s noble in a way, but doomed.
There’s no chance that everything that matters isn’t touched by an LLM in the coming decades, it’s like those refuseniks who didn’t get cell phones because they found the idea creepy (I was one of those, I get it)
I don’t know about noble, it sounds much more like the 5G stuff these days than “no cell phones”. And honestly the “no cell phone” crowd turned out to be right in a lot of very important ways.
Could you name one?
If you narrow, “no cell phones” to “no smartphones”, then the lack of access to social media is probably a net positive.
Which doesn’t even depend on the smartphone.
Twitter and Facebook existed before the iPhone.
Not granting the whole world interrupt priority on my life via a device I’m paying for, for one XD
Not at all. You’re just an avid fan of LLM and disregard any arguments against it’s use as “anti”.
Nah I’m talking about the qualities of the arguments. Me, I’m a pragmatist – pretty much avoided LLMs until 18 months ago cause they were worse than doing things myself. Then they improved enough to use for some things, now they’re good enough to use for a lot of things. And this is coming from someone who’s been messing with ML off and on for a couple of decades. There are some really interesting questions to be resolved in terms of LLMs being good for us or not, but the “anti” crowd doesn’t engage at that level.
Unnecessarily totalizing and reductive attitude
We currently have the situation here where LLMs are being used to find, and potentially fix, code exploits; meanwhile we have cowboy programmers who think they’re Software Engineers claiming AI can’t write code.
AI/LLM can write code, can do so at a Software Engineering level. You can have the LLM write the functions you need, and the test cases to prove the code does as expected. You can review all the above and then decide if it qualifies as adaquate for task.
We also have a lot of irresponsible LLM usage, spewing barely analyzed garbage into the word and causing chaos. We’re using the chaos to judge the entirety of AI/LLM output.
Maybe we should go back to World War II and judge the entirety of computer usage and data processing on the use of IBM/computer equipment by Nazi Germany usage.
I’ve used LLM as tutorial, as advanced equivalent to Stack Overflow, as a code writer, and as a library builder with test cases.
The potential to take thinks to the next level is insane. Burying our heads in the sand isn’t the solution.
Not sure ,but I think tere is a bias when we judge AI tools as useful or not, if you are a good developer/programmer /software related worker, you probably will tell AI coding is just garbage, but if you are Mathematician,not that related to programming type engineer, hobbist or just “normal guy” you are more pro AI coding if you are in the need of use those tools/skills. As for me, I vibe coded some Months ago an analisis tool , an application (Tkinter GUI) that combines two complementary approaches to forecast NAFTRAC’s next trading session high and low: a GARCH(1,1) volatility model and a Random Forest regressor. The final prediction blends both (60% ML, 40% GARCH). Data is sourced from Yahoo
Finance — free, no API key required(not optimal,tho). The tool also builds a historical predicted band across the last 60 sessions using a cached walk-forward backtest, so you can visually compare model predictions against actual price action on every chart load. GARCH stands for Generalized AutoRegressive Conditional Heteroskedasticity. So yeah ,I’m happy with the results so far, would I code some tool like that by myself? sure but it will take me years and many gray hairs(I barely code for Arduino ,the uno )So maybe we just talk about the fair as we are doing in it. by the way i was watching BloombergTV some days ago and one smart guy said we still have 1 or 2 years before the AI bubble colapses, So patience…
Highly interesting, but you are simply seeing a blessing where others see an endless nightmare. The first thing I would say you shoule consider is this; as the complexity of your LLM written code base increases you risk losing track of all of it, how the pieces fit, at this point you could never hire someone to maintain or scale your system up. Having no scalability or cognition of the systems overall design makes business based on your approach risky.
I too can shortcut projects, but we must remember that if you could write your financial systems codebase youd be making good money and not vibe coding.
If I got a masters in mathematics my projects would be a heck of alot cooler than they are, and I could scale a patent or design into a business or career.
There are techniques to mitigate that, also a codebase created by a human team runs into the same problems at a certain scale, so it’s not unique to LLM code.
The main risk IMO is that people think their LLM output is the same as the LLM output produced by an experienced developer – it’s not! A human with experience working with large codebases and large teams knows where the sharp edges are, and more importantly will push back against prompts like “make it faster.” see: grugbrain.dev – most important grug skill is say “no” but that lead to less shiny rock.
+1 ….. if someone writes code with LLM they’d be well advised to try and understand that code and think carefully about the overall architecture if the project’s going to get bigger. There’s currently a big problem with larger projects – as the project gets larger there isn’t enough ‘memory’ to hold it all and the LLM just ‘forgets’ about large swathes of it.
Yeah I’d say you are right on target
I should have said, demo account, Months of monitoring,human eyes on it,logging everything, the ai thing is working nice so far, analysis tool, no trades maker, the balance is the judge :)
People saying “it’s just garbage” probably haven’t actually tried it or reviewed it. It’s not that bad.
Another possibility is that they know they’re lying, but they’re saying it’s garbage anyway for the sake of job security.
This. I’ve been using Claude to write smaller backlog projects I’ve wanted to get done but haven’t had the time or interest to slog through. And it’s done a pretty good job with it all so I can focus on project that are actually interesting.
Wow! Someone cleaned out this place. Even took the lights.
Now i understand your comment …. it’s been troubling me all day :(
Their house, their rules and I even like it. But I’m wondering whether that will manage to get the AI toothpaste put back into the tube.
Still near to daily I’m shocked how many more projects I like are allowing LLM generated code.
SIGH!
Oh absolutely zero percent chance of putting it back in the tube. No way Jose.
I get the argument, I get the hypocrisy of disallowing scraping while allowing LLM-generated code, but the toothpaste stays out of the tube. Scraping will happen, it won’t be stopped just because they say so.
Fundamentally, LLM-generated code was still initiated and perhaps even reviewed by a human, consider it as a tool. Would you ban code written using VS Code instead of Vim?
Frankly, It makes them look like Luddites. LLM-generated code is often cleaner, often faster, detects edge-cases early, writes tests, can test itself and is improving quickly… So I do not get the community decision.
Fortunately, alternative code-hosting platforms exist, but I’m saddened by this mentality as I just switched to Codeberg from GitHub. I have 1 LLM-generated open-source, MIT-licensed, project, which I wanted to make as a learning experience (won’t have any users besides myself probably), that I’m quite happy with… will it be banned?
I suspect we agree on a number of things. I found it relevant to read about the fears people had about the industrial revolution, you might find it as interesting as I did.
Maybe they will use an AI to detect illicit AI generated code. LOL.
Or they’ll do what everyone else already did before LLMs and say “if we catch you stealing art/written work we blacklist you”
Society already knows how to deal with stealing; this isn’t some unknown thing nobody’s dealt with before.
Yep, it’s easy for anybody with a half decent LLM to re-write some humanoid derived code and turn it into something that works better, has better mark down and is generally easier to read and then call it their own. It’s basically theft.
How come nobody does that?
…Does society know how to deal with stealing? Idk if that’s completely cut and dry
The worst part about this situation is the pro-LLM folk insisting its a blanket ban when it just bans stuff exclusively made by AI.
And when you point this out they go “well that’s what it is right now haumpf”.
and when you point out they don’t have to use codeburg they get even angrier because “muh freeze peach”.
ugh
Personally I don’t care about codeburg per se – when I give up on GH (and it’s happening soon TBH) I’m going self-hosted.
That said, “stuff exclusively made by AI” is entirely, 1000%, unenforceable. I add a period to
README.mdmanually – did I avoid ban? Prolly not. So what it’s going to amount to is people spotting AI “tells” in codebases and getting others banned.It’s not really fully enforceable (without watching people type every character at least) but there is a “honour system” aspect to it. And that’s the majority of enforcement for most rules or laws. And this wording at least allows the uses that are basically “enhanced language servers,” fast custom templates*, etc while (in spirit at least) banning “stolen valour” style projects.
They’re also a host without the money the likes of MS, Meta, etc have. Until there has been more legal testing they have to protect themselves. They can not afford a $1.5 billion payout to disgruntled authors like has just happened over pirated training data. It may be apples and oranges but I think the caution is understandable.
*I am not anti-AI so to speak. I thought of these two examples because they’re how I use it.
So I read this article and found it pretty interesting, way more so in the comments, which I always check out. As someone on the outside, I’ve got my take on this whole thing. Basically, it’s about developers writing code and then companies using it without paying the creators. It’s like, “Where’s my money?” Honestly, this isn’t about AI, capitalism, or who’s right or wrong, it’s just about money, plain and simple. My honest thought? Codeberg’s probably gonna end up like GitHub.
There is also the ‘share-alike’ aspect of Open Source. This is how I feel about anything I release to the public. I released it so it would get out there and maybe even be improved by others. If someone uses it for a closed source product that eclipses my open one then that might be harmed.
But if I released it to the public I’m not expecting money.
And if you make money off it.. cool. I’m happy for you. Just make sure to open source your mods. So long as you do I see this as a benefit, not some opportunity to make money that was stolen from me. I despise non-comm licenses and think they are petty.
Anyway, back on topic.. if an AI is going to hand out my code to a closed source developer then that is the same harm as a developer just copying it into their closed project on purpose.
So, it’s not always about money.
Scribes vote to reject work done with printing press
Same energy. Fight it all you want, it’s here to stay and you can embrace it and learn to use it as an additional tool or get left behind.
Life is water, not stone.
Learn to swim, or drown.
I just don’t like when people that want to ban “AI slop” aren’t precise with what they mean.
For example.. I am currently working on a project that has grown into something that I will publish the source soon. I started it as a quick, one-off. It was supposed to be a much smaller project. I thought this would be a good candidate for giving AI a try. So.. my initial code was mostly AI generated.
But.. I did read through it. I did make sure to understand what it was doing, how and why. And I made changes where I didn’t think what the AI did was ideal.
Then.. I had more ideas. As projects do.. it grew. I added some things with the help of AI, again, combing through everything, understanding it and adjusting it.
Then I decided I want to publish this sourcecode. But, I wanted something more organized, not the script that had grown from an initial small hack. So I started re-arranging and reformatting everything.
So.. at what point did this transition from AI to not AI? Was it ever slop and if so when did it stop being so? Should I avoid sharing this project anywhere that says they don’t want AI generated stuff?
Also, I increasingly see comments here on HaD accusing some work or another of being AI and I don’t see it at all. I think some people are starting to see AI in anything they don’t love. I think people should keep that accusation to themselves unless they are REALLY sure. Imagine working hard on something then seeing some 4ssh0les call it AI.