Freesound Forums

Hi friends,
In fact, I think the right question is : which kind of AI are we talking about ?
I mean, if freesound would be able to only allow ‘’AI activities’’ helping sound researchers, let’s say, for analysing environmental sounds to improve our comprehension of animal languages, our to understand how sound pollution might disturb nature and wild life activities for example, it would probably be ethically acceptable.
But, as I mentioned earlier, as Freesound cannot guarantee such ethical uses of the sounds hosted on its website, there should be a way to let users fully block AI activities which, indeed, might as well affect significantly their intellectual property rights.

Wishing you all the best !

Kevin

kevp888 wrote:
Hi friends,
In fact, I think the right question is : which kind of AI are we talking about ?
I mean, if freesound would be able to only allow ‘’AI activities’’ helping sound researchers, let’s say, for analysing environmental sounds to improve our comprehension of animal languages, our to understand how sound pollution might disturb nature and wild life activities for example, it would probably be ethically acceptable.
But, as I mentioned earlier, as Freesound cannot guarantee such ethical uses of the sounds hosted on its website, there should be a way to let users fully block AI activities which, indeed, might as well affect significantly their intellectual property rights.

Wishing you all the best !

Kevin

Hi everyone,

I'm sorry to see some leaving, particularly while we are in the process of discussing the situation, trying to understand each other's position, and also as I was very clear in my previous message that we are open to make changes and adapt. I hope that everyone participating in this conversation is aware about the context of Freesound, where Freesound is being developed, the team that is behind, and what we are trying to do. If that is not the case, I would recommend maybe watching some of the Freesound Day videos. That might clarify some of the positions. We have been always very open to discussions and to incorporate suggestions, always listening and trying to answer as many comments as possible. Freesound is run at a public university, by a very small team of researchers and with minimal resources. We're not big tech or anything like that. We often don't have resources to do many things we would like to.

After that introduction, I'll try to answer some of the things being said since my last message, if I miss something, feel free to highlight it, I'm not trying to avoid discussing anything.

We, as a research team, and me personally, fully agree with the harm that AI can have on our society and that some of you described in previous posts (even though I would argue that it is not only AI that we have to be worried about, and that AI is indeed highlighting other issues already present in our society). We often talk about AI in very general terms, and this does not help our discussions. I like how @kevp888 brought out the question "which kind of AI are we talking about?", because this is also what I was trying to say in my last message. Even though AI can make a lot of harm, there are also other types of AI which are relevant and can benefit the community (and society of course). These are the types of AI that we want to support, from our research group and from Freesound. Of course we don't want to support AI as described in @Milkman1976's post.

The feature we are adding to express "AI preference" is a step towards transparency, but it is definitely not the solution to the problem. We proposed these 3 options, because we think they allow to rule out some unwanted AI usages by more explicitly limiting commercial use (adding to CC-BY-NC), and also by providing the option of only allowing open models. Open models are those that are transparent, that can be inspected and that are accessible to everyone. This is to contribute to digital sovereignty, as opposed to depending only on big tech. We believe this is the way towards building AI alternatives which are good.

The reason why we did not include the "no AI at all" option is, as I mentioned in the previous message, that we thought the previous options were enough to limit bad AI in a meaningful way, and that we don't ideally want to contribute to the "no AI is good at all" narrative. Freesound is a platform built in a research context, very open minded and willing to explore the boundaries of sound technology. Nevertheless, if the community truly believes that the "no AI at all" option is necessary and that it protects you better, we're not against including it.

As you will know, it is not clear legally how generative AI issues need to be addressed. On the one hand we have the copyright issues, for which in Freesound we have the Creative Commons licenses, but these were not specifically designed for AI. On the other hand, there are moral issues which might be partially covered by the terms of use of the site, but that we can't really address beyond that point. The feature we are introducing is a complement to the Creative Commons licenses, but still it is unclear what their legal implications would be. And there's little we can do about it at this point as there's no agreed way of legally expressing the preferences that our feature is introducing (also considering our available resources). But this is definitely a first step, better than nothing. That is why we're adding it. We are sometimes contacted by companies asking if they can use sounds for AI, and we so far we refer them to the interpretation of Creative Commons. Now, we will be able to also refer them to the new user preferences, and, according to our experience with such companies, this will be respected (in many cases at least).

I think we all share things here in the open because we like doing so, and because we believe this is good for the society. It is possible that some of our sounds are misused, or used in ways we don't want. This was already like that before AI, although AI obviosuly makes it worse. I guess the question is also whether we still believe that it is good to keep sharing like that or not. Me, personally, I think that sharing in Freesound is still a great thing to do, and positives outweight the negatives. And even more knowing that Freesound is trying to develop a healthy relationship between AI and the community (which is obvioiusly not easy, as this whole thread shows very clearly). But, of course, other positions are just as valid as this one, and we can also understand them.

I hope this helps bringing some further context to the conversation so we can continue having it in a respectful manner. I can only understand some people saying things like "they don't care", or that "you're selling out" if you don't have enough context about who's the people running Freesound. So hopefully this reads useful to some of you. There's no hidden interest in Freesound that we're not sharing, we're not selling to anyone and there are no external pressures. We're just trying to find a way to constructively benefit from the good parts of AI, and protect ourselves (including all of us) from the bad parts, and that way is not obvious. We're humans and want to remain like that.

And finally, just to address one very specific comment about "why was it necessary to resist the removal of some bots at the time". It is not resisting to it, we simply don't have human resources to fight against that in a reasonable way. You get rid of one bot, and the next day they change the strategy and you have to deal with it again in a different way. We have some ideas for minimizing their impact (and we'll possibily eventually do something), but if we spend the little time that we have deadling with that, then we can't take care of the rest of Freesound stuff. Overall we simply believe there are other things to do which are more beneficial for the community.

Thanks, sincerely, for spending time reading all posts and sharing your opinions.

Thank you, Frederick, for the detailed answer! That's the only place to start. And not with the silent introduction of dubious buttons! Since our world today is overly crowded with such introductions, called "on a voluntary/compulsory basis" this can sometimes cause people to have very violent reactions. Which in itself is probably fine. Because such things contribute to the collapse of rights and freedoms. People are losing more and more confidence in their own kind. There are a lot of scammers around who sow destructiveness (don't take it personally). I will not insist on my point of view and views on AI over and over again. But, personally, I don't want to participate in this in any way. I just want to share my sounds and works here. And enjoy other people's work, because it gives me real pleasure if I find something worthy and interesting. I'll just say about the point where it all started: let the people decide for themselves what is good and what is bad. Add a button that will restore balance and harmony on this site. And I won't care if someone stole my sound, I'll just be calm that I didn't agree with what I didn't want to agree with.

Hi friends,
Frederic, thank you very much for your time and reply !
Indeed, I think it might be important to remind us the philosophy of freesound.org, and as you may have noticed from reading our posts, most of us are committed to the ethical and collaborative aspect of the website, as the team and you.
Personally, as I already had the occasion to tell here and there, I really enjoy sharing my sounds and experience here, as well as discovering the treasures brought by most of the members of our wonderful community.
And as a human being who like to share and interact with human beings, I have to admit that I might be very suspicious with malicious use of AIs (even if, as a blind folk, I know how AI might help people with disability in some circumstances).
So, please don’t hesitate to share the thoughts and questions of the team here, because I can guess most of us care seriously about the future of Freesound and will be glad to help you if you need some advice !

Wishing you all the best !

Kevin

Hello Friends,

First of all, thank you for Frederic's detailed explanation. Thanks also to fellow users for all your input.

I would like to share my views on AI.

I strongly agree with Frederic that freesound.org should strive for the responsible use of sounds by AI, although I also fully understand the criticism.
I may be a dreamer, but for me, AI is a reality that can definitely be put to good use. I would like my sounds to be used with respect for my licence preferences.
Let me give you a concrete example.

1] Bird lovers worldwide

The website https://xeno-canto.org/ contains many observations from all over the world. The data is provided by biologists, but also by volunteers from all over the world. This site helps scientists in their bird research.
I make thousands of sound recordings in nature. I would be happy for an open source AI tool to use all my data to supplement this site with my data. This analyser could then be incorporated into such an AI tool https://birdnet.cornell.edu/

2] People with disabilities

Wouldn't it be great if I could ask my phone (AI tool) to play a sound near a certain location? I could then specify what kind of sound it should be.
Eg. This would be great for visually impaired people or people with tinnitus who want to go to sleep without being bothered by their own sounds.

Long live science, open source software developers, and the fair use of AI.
This will make the world a better place.
My advice: don't be afraid of the rise of the steam train, like people were in the past. (But I already said that earlier.)

All the best to my dear friends around the world.