Why do you expect that? Haiku is not super popular project where everyone is rushing to contribute. Lets see where it will go before making decisions. Banning everything is easy. Making something useful is hard.
Currently AI-sketpics are making much more flood on this forum than people using AI in their projects.
There are limits to what is acceptable. And it’s not just about opinions, as I think Nipos already wrote (even if the tone is indeed not the best).
LLM scrappers are attacking Nipos’ servers, as well as mine, and those of Haiku. This has lead to distraction from the project as we deal with just keeping this place online.
And the deeper impacts like climate change mean people will actually die because of this.
I don’t think that’s possible.
A strict ban on some type of content is not something that we do lightly. Historically the moderation on this forum has been quite relaxed, allowing a lot of things (within reason, I hope). Of course, for “just a forum category” it isn’t so much the case, but Nipos and a few other people are here to push for a complete ban.
This is justified by the extreme consequences of LLMs, wether it is direct ones to Haiku developers, sysadmins, and anyone trying to host a server on the internet these days, or on the more global climate change side, for example.
These rather important consequences are what may justify a ban of LLM content. Sure, some people will say that this is a political choice and that politics should not be talked about here (and they already did). Some will say that it’s just opinions. Some will say they don’t care. Some will say they think their usage of LLMs won’t really matter in the grand scheme of things. And, you know, maybe they’re right. Maybe banning LLMs from one forum would be just a political message and have no practical effect. That’s quite likely, in fact.
But still, the issue with LLMs seems so important, so impactful, that, to me, sending this political message seems important. And the fact that we’re even considering adding a category to the forum means the debate is built on the wrong basis: that the ban of LLMs would be on the grounds of low code quality, low-effort, high volume slop. Which is not the case for two reasons: I think the other issues are more important, and also, there are people who do advocate for some usage of LLMs while still reviewing the code and keeping things in control of a human. By discussing this side of things, you have already decided to ignore the first level of issues.
If you vibe code an app on an LLM that you have trained locally, with low resource usage, not contributing massively to climate change or even local warming, and without DDoSing my or other’s website? I’m fine with that, you do whatever you want, and in that case, the question of making a category to ignore it is valid. But with the current LLMs? We shouldn’t even be having this discussion.
I hope you can see how and why this is (possibly) irreconciliable, and why some people insist on keeping the discussion on that level, and talk about the why? I want to make it super clear that people voting to allow LLM usage have thought about the fact that this is causing a DDoS of our servers, making climate change worse, and generally making the world a worse place in very real ways. For me, this cannot possibly be ignored, and I don’t want to platform people who think otherwise by giving them a space in this forum. How can I then have a discussion on “what to do?”.
Will LLM scrappers stop attacking or reduce attack rate if Haiku will ban mentioning AI-tools assisted project on forum? Definitely no. It is a kind of cargo cult to think that it helps or matter in any way.
To make LLM scrappers stop attacking you need to add robots.txt and protection like Anubis, maybe make login required.
That is of course true. But even if people’s views cross that limit, or are factually untrue, we still have to deal with it somehow. That’s what I’m advocating for. We need some kind of process that in the end produces guidelines for dealing with AI within the Haiku project.
I’m very much aware of the climate change impacts and I said so in several other threads. But I deliberately chose to not state my own opinions this time in the hope we can find a formal process to deal with the AI situation.
I greatly appreciate the acute environmental awareness shown here.
Why then, has there been a bitcoin mining app in the repository since 2023? Doesn’t anyone realise how bitcoin server farms are affecting global warming?
I’m sure it will be removed forthwith. We wouldn’t want to give the impression that we are only environmentalist when it suits us.
This position unfortunately requires a fundamental choice. It doesn’t leave much room for any compromises.
I feel you, but I don’t think we should or even can go down that route as a project.
I think we have to be more pragmatic about this topic. We simply cannot control AI usage outside of our own doing. We can have a policy of not accepting AI generated code into Haiku’s own codebase, and we already have that. We cannot control Mesa or any other software project we depend on. We cannot do everything ourselves, we are and we will heavily build on top of other open source projects outside of our control.
I think that LLMs are mostly trendy right now. Most serious developers will come to the conclusion that they don’t really help much in developing quality code. I’m afraid, though, they will stay until something better comes along.
I would personally welcome a text on our website detailing our reasons not to use AI, on a prominent position, not politically neutral. But I don’t think banning AI from anywhere beyond our own code will work or do us any good as a project.
Well, this is not true at all because haiku is by far one of the most developed lesser known OS is more developed than any OS, not to mention, it has the most mature drivers out of serenityOS and reactOS not to mention people wouldn’t see haiku as a AI sandbox, they would see their successor BeOS also why did you use AI to reply to this? This is an open ended, answered that comes all answers. The fact that you needed AI to do this does not really reflect a good source of argument or using the one thing you trying to say is bad also doesn’t reflect a good source of arguments
Till this day haiku literally managed to complete all of its goals, and still make an identity for itself and still be the BeOS we all love and know today regardless if they include AI or not haiku is one of the few OSes that you can truly daily drive even you’re used to Linux or windows
Lastly, windows widely accept AI and its identity didn’t even change. Most of things. People don’t like is literally the fact that it’s been bringing out bad updates. Linux does not widely accept AI in fact, they said AI can be used, but it must be highly moderated yet that still does not change the fact that they are still the oasis of the mainstream world. They are not looked at because of AI, but because they still have a strong identity Haiku is one of those few OS that you can proudly say that can’t change regardless if it does include AI or not
Pleaseeeeeeee don’t post LLM sludge. This is a forum for people.
honestly, my view of this has changed. I went there to still be a hard ban on AI, but I still want there to still be used for it and just making software with people or drivers so we don’t have to constantly rely on FreeBSD not that there’s nothing wrong with that, but even BSD mainly takes most of each drivers from Linux and honestly I don’t wanna have to constantly look at Linux slop in haiku
Fair enough; in that larger context, you’re absolutely correct.
it made me easier and faster to response; serenityOS is for VM only as author mentioned; ReactOS is targeting Windows API compatibility, that’s a long track targeting billions of driver and is only XP now ; Haiku doesn’t have settled 3rd party drivers suitable for modern computing, its behind FreeBSD and that itself is behind Mac, Windows and lastly Linux, Really cant say much there becouse of my BOOTCAMP macbook, but even he would love some Haiku, since is very old, but firmware and loader would need some work in the middle work specialy for macbooks.
Talking about response, yes, botted-like response in AI is NOT.
This I disagree with: I think Haiku could very well make a “no generative AI material on any Haiku Inc resource”.
This would definitely mean only using older versions/potentially forking some underlying open source resources, but it can be done.
I would personally agree with that stance; it’s the same one I have for my own systems and why I want Haiku to at least require labeling here: so that I can personally maintain my 100% no-LLM-generated-code stance.
As long as we’re having the larger discussion. 
Still doesn’t make sense on why you didn’t take time to write out your own idea for a open ended answer and question, let alone using AI to prove why ai is bad for haiku that a circular argument.
Actually haiku driver can be used on Linux and FreeBSD via hyclone, haiku has it own driver and has the latest FreeBSD driver along with this haiku is still pushing strong with it hardware support and firmware support that it still has a faster UI than window and Linux, but overall haiku has made a name for itself like Linux in the middle of it life or window 10 in it prime which is why haiku regardless if ai is included or not mainly because people use it for it fast UI and BeOS compatibility
No, it can’t. HyClone is only for userland Haiku software, it do not support Haiku kernel drivers.
That’s unfortunately true,but also not exactly what I see as a huge annoyance.
Let’s just hope that the big longstanding projects that Haiku depends on and some big ports like IceWeasel still have some form of quality control.
That doesn’t mean we can’t ban advertising of low-quality slopcoded projects from the forum.
I wasn’t going to chime in, but clearly people take “discussion of AI is fine” to mean “it’s okay to post AI slop without even reading what real people took the time to write”. And “compromise” always seems to mean “accept the bad stuff without comment; don’t ask for any concession to the good”.
Consider this: NO.
Yes, I actually agree with that.
The “stay away from politics” rule was put in place, I think, precisely for that reason. Such discussions are annoying and draining energy for everyone involved, and since there isn’t room for compromises, they don’t lead anywhere or are destructive.
And I should not be allowed to do that.
It puts me in a difficult position. But I will make my own compromises. I know people who are vegetarian or vegan because they think killing other animals (that they consider as equal to humans) for their own feeding or enjoyment of eating is unacceptable; or even just for environmental/ecological reasons. Yet they can be friends and even share a table with people who eat meat. There can be some people making compromises such as eatling less meat but still eating some (makes sense if you are doing it for ecological reasons, maybe less so if it’s the killing of animals that bothers you). And, maybe more relevant to this discussion, this isn’t something that one would impose on an entire community either, and even then, doing so on a relatively small place like Haiku wouldn’t change much in the grand scheme of things.
So, I think that’s where I stand now, I just wanted to say “wtf, you all use LLMs?” just as a vegan could say “wtf, you eat dead animals?”. It is indeed a personal choice, and making it a more general one at the level of the Haiku project requires abandoning other things such as “stay away from politics”, a rather big shift that I don’t even know if it would be right. This is something that, until now, we have decided to not resolve. Hence the ban on LLM generated code in Haiku for different reasons, for example, to not have to talk about this.
Obviously the right thing to do when vegans and meat eaters share a table, is not talk about it, and pick some other conversation topic. So far we have all failed to do that here.
I also make compromises in other places in my life, and in some cases I don’t even hold myself to the same standards that I demand from Haiku. For example, I always insist that Haiku should only use open source software for its infrastructure (and, for example, not use Github). But on the other hand, I use closed source software in other places, and I even write closed source software in my paid job. Some people would find that unacceptable as well. Indeed some people find me very annoying for not wanting to use Github, and at the same time some people think I don’t do enough to push for that. I guess eventually I will find my balance for LLM things as well, but right now it is quite difficult.
Respectfully, in my experience it’s the meat eaters who invariably lash out at you for even saying that you’re trying to eat less meat for whatever reason. Not necessarily ethical, though if you do invoke ethical arguments, then they’ll really get nasty about it. Much like Rust fans (thank you, Michel) or people who mock biking and electric vehicles. Or, nowadays, generative AI users.
And yes, I also used to write proprietary software for a living. Had to make a living after all (it’s always the same tired excuse). In the end I only made the world a worse place and burned out, partly for this reason.
That’s how I got radicalized.
No, it can’t be done. We don’t have enough resources to ship a fully working OS with a fully working browser, so we don’t have enough resources to do that AND ALSO fully maintain MORE third party applications.
I don’t have much sympathy for AI but when I read this kind of arguments it makes my guts twist.