What is Ardour's Generative AI policy?

Okay, I see that using LLM’s for generating code is banned. Does this mean that LLM based functionality is not? The policy does not appear to cover LLM based functionality (the MCP server).

1 Like

Our policy re: LLMs is based primarily on its impact on the project itself, despite the two main developers’ severe concerns with LLMs in general. At this time, we are not seeking to make a statement about LLMs by banning (or ignoring) any possible interactions between Ardour and an LLM (that could change). We are, however, seeking to avoid the problem of random individuals overwhelming our code review process with LLM-generated submissions, along with the issue of incorporating code that has subtle but deep misconceptions making its way into our codebase.

So at this time, “LLM-related functionality” is acceptable.

3 Likes

I earnestly and respectfully ask you to consider banning anything to do with “generative” AI outright. It’s issues are quite foundational and unlikely to do anything but worsen with time, such as it’s environmental impact, it’s effect on the price of hardware, and that’s not mentioning its ties to the war industry as well.

I’ve been using Ardour for around a year, supporting it financially since I made the switch and have been getting great use out of it. I would have tried to contribute directly as well, if I could code worth a damn.
I’m proud of the fact that my audio workflow is entirely FOSS and AI free (with the extremely troubling exception of the Linux kernel itself now). I’ve recommended it to others using that post of mine mentioned by OP, as I was so happy with your answer and knew it would put others at ease, as well. It’s kind of funny, because I soon realized how accusatory I came off and regretted my phrasing, especially given your response. I still wish I had chosen better words, but I even used that post to suggest Ardour for the Starlight Network’s No-AI list! I was oddly proud to be the one who got it there too, it simply meant that much to me.

I appreciate the fact that Robin and yourself (I assume that’s what you mean by “the two main developers”) are concerned about “gen” AI as well, but I urge you to let that bleed into Ardour’s policies for more then just copyright or quality reasons. It’s not unprofessional and would be quite a good look for the project, in fact. Heck, even if I wasn’t the sound nerd I am, I’d still throw a few dollars at the project every month just for that. Backing such projects is even more important as open source software grows complacent with these slop machines, so long as they’re invisible to the average onlooker.

Apologies for the lengthy message, though I hope it communicates the importance of this topic. Having switched to Linux from Windows in the first place to escape corporate theft devices, I thought I was retreating to a place predisposed against such things. Now I find myself becoming a staunch advocate for strict anti-AI policies almost out of necessity. If you want an example to work off of, I can suggest the one adopted by elementary OS. I’ve been considering installing it on my music creation rig, as I want it to be close to AI-free as possible.
I’m deeply grateful for the work you have done over the past couple decades to bring Ardour into the world, continuing to grow and maintain it all this time. I’m also grateful for the kindness you continue to show myself and others who are learning, no matter how basic the questions may be, and for your continued positive response to user feedback. The fact that you finally added a separate MIDI window for those who demanded it, a Tool-releasing-Fear-Inoculum moment, speaks to that. I know it means you’ll consider this as well, and (I’m not sure how to say this without it sounding like a passive threat, but) I hope you’ll make the right decision.

Thank you for reading,
cloudskater <3

8 Likes

AI seems to be here, and here to stay.
Not being a SW developper myself, I cannot rate if this a good or bad thing - except that the code quality might be poor and hard to test - or if it is the fear everytime a new technology is introduced (which is a very human behaviour of course).

For me it looks like SW generation now has to focus more on the specifation phase which I believe is a good thing.

Anyway I’m very curious to hear pros and cons of AI supported SW development.

Pro: if you know what you’re doing, AI can be a helpful tool to accelerate the tasks you’d be doing by hand otherwise. It can free a capable developer of mundane, repetative or boring tasks, and free time to focus on the more important tasks.

Cons: If you don’t know what you’re or the AI is doing, you have a bullshit generator that’s good at lying market itself as the best thing since sliced bread.

AI is a capable tool with lots of good use cases. Unfortunately, it’s overhyped beyond the moon and forced into a lot of use cases where it really does not belong, or forced to do tasks it is not ready to do.

Not into programming myself, it is at least imaginable quite a problem is getting flooded with “code”.
Like it is a problem in others and pretty much any topic. You get a wall of “info” in the blink of an eye, some things correct, some half-wrong and some fully and even laughably wrong. Before you have that sorted, before you even started to sort it, assuming you care for it, you already have 5 more walls of info.

(Not much interest in general, i let an llm write me in 5 seconds a lua script for ardour. Impressive indeed. A pity it didn’t work at all. And the same happens in forums and chatrooms. You ask something, someone gives you the llm generated code in the blink of an eye, but it doesn’t work ).

Perhaps that rather social problem, simply gettting flooded with half-wrong “info” ( or code) will diminish over time. It sure is annoying as of now.

Generative AI is a theft machine.
I’m not sure how it handles coding assignments, but when it comes to art, it simply snatches and assembles pieces it finds online. Use it long enough and you’ll get the idea. Anything AI related where you interact with it’s server, run like hell. Being completely offline is your best bet :slight_smile: .
I hate to even talk about it (makes you look paranoid etc), but i cought it snatching something mine on two occasions. Once it found my classical guitar intro i did for a student movie back from 2005, keept the notes/harmony/tempo and changed it to piano, and put it in an indie horror game trailer. It’s very unlikely that it’s coincidence because the chords/positions aren’t standard ones, and it’s created by ear. Second time it snatched the bass line+rythm, tempo and the title of an electronic track, and put it in a pop-sounding tune which ended up on artlist. Now, this wouldn’t be even weird (the original idea ain’t that off) only if the title wasn’t snatched also :), that’s simply too much coincidence .
You have Ola Englund’s video on youtube where he shows what AI snatched from his tunes.
AI isn’t “creating” anything, that’s an ilussion a lot of people don’t have the courage to admit, because it means most powerful men on earth are outright liars and thieves - which is true, in opinion. It’s just to much to take in, nobody wants to see that kind of sight all around them.

4 Likes

I sure understand AI is a threat in creative aspects

But, keeping the discussion to SW generation
remains the question how the generated code must be classified inline with the copyright restrictions of the training data if it is legally acquired.

And if it is and the copyright restrictions allow I do not see a problem in reusing code in whatever form be it by human or artificial intelligence.

Thinking about it even for any other type of AI applications the real problem is not AI but the illegal use of training data. (I think there was a case in Germany about exactly this last year)

AI should be a tool to support humans, help us, make us better, not replace us, separate us, isolate us…

Unfortunately, it can do both and will continue to do both for a while…

2 Likes

I think good use cases exist, at least in the sense of searching for things, with different possible objectives behind the searching.


While Python is not the same as Ardour, I’m slightly reminded of the following Python post and chain of replies upon replies, which I was following daily for a while when it was active, though now I wouldn’t recommend reading all of it, for one can probably make better use of their time.

Here’s 3 quotes in that topic, from someone who’s a CPython core developer, for the purpose of perspective I thought was worth sharing.

“…I want responsible use. AI is a new tool, but brings new potential benefits as well as new potential dangers. …”

…Because the PSF can only accept contributions that the contributor has the legal right to license, all submitted material must be free of copyright or licensing conflicts, including material produced with the assistance of AI tools. Contributors are responsible for ensuring that their submissions meet this requirement. …

"…we already have a policy. “Keep the status quo exactly as is” is a third proposal, and wins by default. …


(For clarity, I’m against building datacenters though. At least in the current sense of noise pollution and consequences for local electricity bills. Not gonna spend time thinking of every reason to despise them.)

That’s why I am very strictly against any “generative” AI, but not machine learning in general. I could be wrong, and maybe all LLMs are unethical by design, but to the best of my understanding, machine learning can be used to help predict things, like spotting medical issues before they become life-threatening, or at least pointing doctors in a general direction.
However, due to the fact that all AI currently marketed as such is built on the same foundation that disrespects artists and people as a whole, I see little indication that these distinctions matter as of right now. Once we get out of this bubble, maybe earnest people will be able to develop such technology for good, but it will be a long time if so, and whatever comes out of it will only be a distant cousin to “generative” AI at best, because “generative” AI is built from the ground up to do harm.

2 Likes

Software is a creative field in and of itself, and generating code is no better then audio or images. I respect @x42 and @paul not only for the quality of Ardour as it stands, but the continual effort to make it better and listen to user feedback. Those two, as well as contributors who’ve come and gone, all leave their mark on such a project in their decisions, ideas, how those ideas are implemented and optimized, everything. Programming is an art form, one I appreciate every day as I continue to suck at it!
Having only recently learned what Ardour mean, I think it lives up to its name. You can tell its made with love when it gives you a silly plugin to turn your keyboard into a damn telephone, and when you look to the description for a reason why, you get an ET reference. “Gen” “AI” might be able to licit confusion and much worse, but it’ll never be charming like that, let alone produce anything worthwhile.

1 Like

I was curious about YouTube, so i did a quick google search. It only took one question, the info is not even hidden, yet, so many people are not even aware of it. To fools this looks like world of opportunities, freedom, progress. Guess i should have known all along… It was so logical & obvious.
And now what? Go ahead, be an damn artist, be an “content” creator, create something unique, waste more of your life :slight_smile: .

So that appears to be an AI summary response, so while it may be correct, I would look for a better reference as well as to what their policy is if you are trying to determine this.

Then again what Youtube does or doesn’t do isn’t necessarily on topic for this honestly.

  Seablade

Indeed, and not just limited to medical, but that opens a different can of worms. In your example it could turn doctors into an accountability sink. A key difference is if you choose to use a AI as another tool in your set, or if it is forced upon you. The latter is indeed something where I would subscribe to your boycott of AI, but in general I don’t agree with a ban of generative AI per se.

-=-

Anyway back to audio: There was a nice presentation by Fares Schulz here at the Berlin Audio Dev meetup with nice overview:

In fact the first 8 slides are a nice history of Neural Music Production since the the late '50s: Advances in Neural Music Production

Some of the approaches are very creative themselves!

1 Like

I have lately bitten the bullet and familiarized myself with AI deeply.

AI is just a new clever way of grouping close together concepts and things that belong together. When AI is asked a question it will be split the question into coordinates that map to those concept groups in AI’s memory and this is how it is able to produce text about those things. The same goes with music and pictures and videos, it predicts notes, sounds, pixels, etc.

It is just mechanics and a very clever human invention of how to group knowledge in a new way that makes that knowledge much much more usable.

Previously I felt threatened by AI because it might take away the things I love: music and coding. After taking time to learn AI I feel that it is just a tool for humans to use and it lowers the threshold for entering a field. I can compose a song and use AI just as a group of talented studio musicians to realize my vision of it. I tried it and it works very well. It is still my composition and lyrics and only “played and sung” by AI. Discovered that process from this video.

If you remember when punk - music was born (if you are old enough :slight_smile: ). It lowered the threshold for common people to enter music, since it kind of gave you the permission to make songs even when you weren’t great at playing an instrument or singing. The same happened with rap, it was easier to enter music because one didn’t need to go through the long process of learning to play and understand music. Now AI is doing the same and lowers the threshold entering music.

AI is just a tool to help anyone achieve a goal without a long learning process. You can use it where your skills lack to fill in the gap and help you be more creative. For example my singing is poor but I can play a couple of instruments. So I’m planning to only use AI by giving it my vocal track to make it pitch perfect and record everything else myself.

3 Likes

While I am not thrilled at all about generative AI I also faced the fact it is here to stay and to be ignorant of it especially as a person in late middle age could definitely put you at a disadvantage in the future in too many ways to fathom. I don’t need AI to make music or graphic art but I am not very good at coding and as someone who has never been great with advanced mathematics I really don’t think I have the knack for it but I do know exactly what I’m looking for in a technical sense. AV Linux is full of script snippets to do various things so I decided to bring these ‘napkin sketch’ utilities to AI and (a) improve them and (b) learn in detail about what steps were involved in the improvements so I could at least read and understand the code better. To be honest I can’t really describe it as easy, even when I had a very defined idea of what I wanted the AI (Google Gemini 3) would often need 5 or more iterations to get to a fully operational running script, if things were not going well it would panic-add redundant features and over-engineer things and I often had to tell it to start again from an early iteration. Even when I told it specific details like icon names and locations it would forget them occasionally and it would undo some of my script edits even when asked to retain them in new iterations. Even though I was signed in for it to keep a ‘memory’ of our previous conversations and projects it would often need reminding and wouldn’t carry relevant ideas into newer scripts unless asked. I felt like I had to essentially re-introduce myself and what my goals were at every new session. This all sounds like it was inept and useless but this is not the case, it just needed a lot of detail and the onus to be organized and keep track of the fine details was definitely my job, as was the in-depth testing. In the end I got a really great set of utilities and although a lot of time and testing was required I ended up with something that I definitely couldn’t have done unassisted. I also learned a lot about Gemini’s current strengths and weaknesses which will make me more effective at prompting it in the future. To get unlimited help with a bash script or the expertise to fix a stubborn theme css file without asking or bothering a human is something I think I will find great value in in the future especially as a spare time Distro maintainer. Overall I’m very glad I set aside my prejudices (and fears) and saw firsthand what using AI is like.

2 Likes

Good point/parallel, but as a self-proclaimed punk rocker, this dosen’t sit right with me.
I listen to various derivates of punk since i was in 3rd grade of elementary school. I like things like The Ruts, Dead Kennedyes, Bad Brains, Ramones of course, The Buzzcocks, PTTB, The Blood, The Damned, Agent Orange and various variants forms of artsy or HC punk. You could say punk rock is my home (aldo i do listen to mostly anything, i don’t realy consider myself genre-centric).
The whole point of that movement was to do it yourself, and the rest was in the hands of faith. Not the machine, not the robot…you. Best to your ability, so it often becomes cute, commical, annoying or straight up awful :slight_smile: .
Look, i used AI already in various ways - used Suno Studio, used AI vo, AI video creation tools etc. I might be forced to use it even more in the future, and when commercial aspect is in question, i don’t mind/care, but i simply don’t see the point in using it in my own creations.
If it comes up with the melody or part of it, that’s not my melody anymore. If it rearanges my arangement, that’s not my song anymore. If it rocks on drums and make the song shine, that’s not realy my recording anymore. And the inevitable part - it stores all of that “colaboration” on a server, are reuse it when prompted by someone else. The whole point of creating something, especialy if you strive for uniqness, becomes questionable. It’s all becomes watered down, stripped of esence, meaningles, an “okay, whatever”.
And on top of that, it’s literaly steals from contemporary artists, and we should all accept that as normal. I’m not going to pretend i’m okay with something like that.

5 Likes

Your point about accountability is fair. I’m not here to argue for AI, obviously. The medical side of it isn’t something for me to figure out I just hope that whatever genuine use it does have, the “AI” they use isn’t the “AI” we have now, and is actually ethical.

I digress.

I don’t see how AI being forced or used at will makes a difference in this case. My points about gen AI being unethical still apply across the board, which I say not to be rude, I just don’t see how any of us should accept “gen” AI into Ardour in any form.
The whole concept has only gotten this far because people roll over, saying we must “get used to it” or be left behind. I apologize for the extreme example but… you know what they say about a Nazi bar? Software works similarly, and we best keep tech bro “AI” nonsense far away from open source, as it goes against the whole concept of FOSS in the first place. FOSS is about lifting each other up and trying to create things by people, for people, with the help of other people. “Generative” AI is the antithesis of that, and I don’t really want to ramble more because I’ll end up repeating myself.

If you are concerned about AI, please, please keep Ardour completely free of it, in code and functionality. It would give us much more faith in the project for this action to be taken, and honestly, it’d simply make a lot of people happy to see from their favorite DAW. <3

As someone who loves punk music and all it stands for, it should be obvious that “gen” AI spits in its face. If you think a community that welcomes inexperienced musicians to learn and grow, not mocking them for their lack of form or knowledge, is comparable to asking maliciously crafted software for audio built from or assisted by metric tons of stolen art, I don’t know what to say.

"If you remember when punk - music was born (if you are old enough :slight_smile: ). It lowered the threshold for common people to enter music, since it kind of gave you the permission to make songs even when you weren’t great at playing an instrument or singing. The same happened with rap, it was easier to enter music because one didn’t need to go through the long process of learning to play and understand music. Now AI is doing the same and lowers the threshold entering music.

AI is just a tool to help anyone achieve a goal without a long learning process."

Where is the joy in learning or creating anything? I’ve had a lot of trouble with the education system over the years, and I’m sure I’m making no headlines by saying it needs copious reform (and should be free to all), but if I had decided to farm it out to a plagiarism machines even part of the time, not only would all my “work” have been invalidated, but I would have learned nothing, and I would have wasted hours upon hours of my professor’s lives with slop I didn’t even create.

I’m trying to remain calm, but the idea that the learning process for music, visual art, code, whatever, is something to be skipped over rather then relished in is an insult to all creatives.

2 Likes