I know that AI-assisted development — and especially “vibe coding” — has been a rather contentious subject around here lately. I’m not posting this to reopen that debate, or to suggest that this is how Haiku itself should be developed. But since the origin of the code matters to quite a few people here, I’d rather be completely upfront about it: haikode is heavily vibe-coded.
I originally made it for myself because I wanted an AI coding agent I could run directly on Haiku while debugging things and working on drivers, without having to keep another machine involved.
It runs natively on Haiku, talks directly to the model providers over HTTPS, and is written using only the Python 3 standard library — no Node, Bun, pip packages or remote server required. It has a curses TUI, a plain terminal interface and a native BeAPI desktop frontend.
So yes: I’ve been using AI to build something that helps me use AI to work on Haiku, which is perhaps a little recursive. And I’m fairly sure I’m not the first person to call a project “haikode”.
My favourite feature is probably one that serves absolutely no practical purpose: the haiku.
When haikode starts, it greets you with a haiku from its collection. When you finish a session with /farewell, the agent writes a new haiku based on the conversation you just had, signs it with its model name, and saves it to your personal collection. Those farewells then become part of the pool of poems that can greet you in future sessions.
Since Haiku itself is named after a form of poetry, it felt like the application ought to behave accordingly.
It has become genuinely useful for my own Haiku work, particularly when debugging and experimenting with driver development, so I thought others might find it interesting too.
I honestly don’t understand why, when communicating one would feel the need to summarise with AI, a post that they had supposedly just written.
What purpose could that possibly serve, and how could someone be so deficient as to be unable to summarise (by themselves) a project they had just been working on?
This whole reliance on AI thing really makes no sense to me and is frankly worrying that such a huge dependency has developed among so many people in such a short span of time.
I have colleagues now who have an AI schedule their day for them and then they are surprised sometimes when they look at their own calendar app, like… oh, I have a meeting with ____, best head off then… How much control over their lives are people willing to cede to other people they’ve never met?
It is disrespectful to call people deficient because of using specific tools. People are free to use any tools they want, make personal schedules with AI etc.. Nobody force to use that tools yourself, but you have no rights to enforce or prohibit using legal tools to others.
Something being “legal” doesn’t automatically make it morally correct. If a person has to use a tool with a really bad environmental (energy and water) and societal impact (displacing people to build datacenters, taking jobs) that was built using stolen assets (these companies routinely ignore copyright) and slave labour (most of the electronics industry) instead of writing a bit about what they created, then that person loses a lot of respect in my eyes.
There are no universal morals, many people genuinely think that using AI tools are perfectly morally valid. It is disrespectful that you come to forum topic about AI tool and start talking about your moral concerns about AI tools. It is your personal problem problem who you choose to respect and who not. Coming to every AI topic to show your disrespect is just flood. There are already multiple dedicated topics to discuss AI concerns.
Please tell me: In which moral framework are children working in mines alright?
Coming to every AI topic to show your disrespect is just flood.
I’m not doing that? I think you have me confused with someone.
All that being said, I agree with @GraXXoR here. Why would you need AI to write a small post about what you created? I personally enjoy telling people about what I made and wouldn’t want that to be taken away by a machine.
Yeah, this went more or less as I expected, unfortunately. But “message received”, loud and clear.
I’ve tried to edit my original post, but I am unable to, due to “slow mode” being enabled. I have considered deleting the post altogether, but for the same reasons I cannot. Bosses; if you want to delete it, be my guests. As I said; I made this for myself because I needed it for my own work. I only wanted to inform others in search of the same thing.
To whom it may concern: I respect the fact that your opinion differs from mine. I am sorry if I have offended you in any way, it was never my intention. Nor was it my intention to cause harm on “children working in mines”.
If you want to have even more fun, implement the ability to connect to an MCP server. For an example on Haiku, search for “Pippo MCP” on Codeberg—I’ve been playing around with it for a couple of weeks, and it integrates really well with Haiku, both for BFS and for interacting with applications.
Yeah, it is disrespectful. It is the official position of the forum policies that you shouldn’t be disrespectful of other users, so in general let’s stay away from emotionally-loaded characterizations of other users’ choices here, please.
Doing this, however, is not disrespectful; it’s just disagreement.
As long as the topic is in the “Proprietary & Other” category, though, it would be off-topic; this category specifically allows generative AI material and is not the place to debate the morality or general desirability.
Edit: To state what I meant more clearly, it’s off-topic for the thread, which is about a specific application; debating the morals in general wouldn’t be off-topic for the category. Sorry to be unclear.
Allowing talking about AI generated software is not the same as forbidding people from talking about it…? Users are free to voice their dissent, or confusion about this sort of thing.
@x512: So far you have been the only one coming to every AI topic to moralize to other people that they should just suck it up, and some people think it’s “a-ok” and therefore they should not be allowed to talk about it. That behaviour is certainly not ok on this forum. People are allowed to discuss this stuff, even in the AI category. It’s not a blanket check for this, it just means you shouldn’t put AI stuff into unlabeled topics. (The slow mode is set on these topics for this reason…)
@bighill You did violate the forum rule once, yes, but the purpose of the rule is not to have an excuse to kick people out for using an LLM; as long as your posts are written by yourself in the future one mistake should not be a problem; No need to delete the post.
An AI tool has its place, and what that place is can only be understood by understanding what the tool can and cannot do. Thinking that AI is a competitor to the human mind is foolish. If we talk about copyright infringement, then yes, AI violates copyright because it secretly uses material without the authors’ permission. For this reason alone, AI can only be used for research and play.
Sorry; I really didn’t express well what I meant there. I can only plead overwork.
I meant that debating the morals/ethics of LLM use should be considered off-topic in a thread about a specific use of LLMs. If it’s not, then the same arguments will keep happening on every thread that mentions LLMs. If we’re going to allow posting about LLM-based projects, it feels unfair to not let the LLM users be able to discuss their projects among themselves in peace?
Ugh; I’m still not expressing what I mean very well; that probably means I need to think about it more too.
Some poeple chose to use unethical, polluting, whatever-the-problem-is tools. Some other people are not willing to ignore that problem. The fact that it keeps happening does not solve the issue.
Why should Haiku, a project that has banned LLM contributions, provide a space for LLM users to discuss things in peace? If that’s what they want, they can create their own forum. But they chose to post here, so I assume this is not what they want.
Then maybe we shouldn’t allow it at all. If you want to keep things neutral, either you allow people to talk about their pro and anti-LLM opinons, or you allow neither.
Everyone, please calm down first. As far as I understand, it has been an agreed-upon principle in the past that code generated by LLMs should not be incorporated into the main Haiku code for various reasons (such as AI slop and copyright issues), as far as I know. And I think this reason is reasonable.
However, I’m not sure whether the principle of banning applications made with LLMs or even writing that received LLM assistance is a fixed rule. If there was something I didn’t know, please let me know.
From what I’ve read in this discussion thread, a kind piece polished by an LLM seems better than a human-made piece that is unfriendly to others. In any case, if you want to consider it “cheating” whenever an LLM is involved, would you also consider a child’s writing, written with a mother’s help, as “cheating”? AI as a technology can certainly be criticized, but I wish it would be a little more lenient toward the new people who come to the community.
Apologies for my language earlier. It was unprofessional…
My personal stance is that AI should be kept well away from all hobbies and anything that is ostensibly a human-lead community. Regardless whether or not it is considered “cheating”, sustained exposure to AI assistants clearly, scientifically and demonstrably have measurable negative impacts on human intellect, memory retention and recall, interpersonal skills, mood, creativity, not to mention the financial, environmental and social costs that are destroying communities across the US and beyond, bringing the water and power grids to their knees wherever a data centre springs up, forcing up utility prices and destroying land value for miles around.
The devastating impact it has had on the home computing and electronics markets as a whole should also not be understated. And more personally, I think any technology that appears to be responsible for assisting in multiple suicides should be reined in.
IMHO, preferably it should be limited to when absolutely necessary to protect one’s job, for example, where demanded by a supervisor or boss. So I agree on Haiku’s current blanket ban on AI and believe it should be extended to all media and documentation, too. Again, not because it’s cheating but because of the overall negative effect that sustained AI use has on those who use it extensively.
Anyway, once again, sorry for the tone of the comment, eariler. I’ll be saying nothing more on the topic here on out.
Cheating is only applicable when code is written with educational or training purposes, especially when submitted to some kind of school/college/university etc.. It is not applicable to production code.
Cheating is only applicable when code is written with educational or training purposes, especially when submitted to some kind of school/college/university etc.. It is not applicable to production code.
It would also be cheating if you tried to submit AI generated co to a project that does not allow it while trying to pass it off as human-made.
Regarding OP breaking the rules by using AI to write their post, it was not made clear here, but it was in the AI friendly forum - OP used AI for translation help because English was not their first language. This seems like a very legitimate usage of AI to help with forum post creation, even if technically it is still breaking the rules.
