Implications of AI generated content

I see nothing against crediting the bot as the artist. Just the human needs some credit for coming up with the seed phrase and decision to start the process. Producer? Arranger? Initial Idea?

For now, in most cases, someone needs to press the initial button. And there should be space to put them somewhere into the chain of credits. They will name their output “Charlie’s Magic Band” or whatever they feel like. And that should be how we credit it. Where we know who is behind it, they need a credit somewhere.

Ditto on the Work side. There is a level of “collaboration” here. The algorithm did the real work, but it from an idea. As to the dataset? That would be a “based on” relationship.

And no, the devs of the algorithm do not get credited. We do not credit the person who constructed the Synth for tunes made with their instrument.

It is my understanding with these AI tools that the same prompt will not always produce the same result; potentially not even a very similar one. That to me makes its function not that of an instrument. So I would go so far as to say that the entity providing the prompt deserves very little credit in most cases.

As a separate thought, I do not feel the works used for the AI’s training set should be ignored; any pre-existing work used to generate an output should, to my mind, be credited.


I never came across anything like that but, IMO, it’s more simple to completely ignore that, for the time being.

Some additional thoughts…

Prompting definitely affects the output, but the output with many algorithms (such as stable diffusion) are non-deterministic, meaning the same prompt will generate a different output if repeated. Also worth nothing I think is the fact that the AI is generating the output based on training from millions of existing works; this is synthetically “standing on the shoulders of giants” at scale, and I wonder how different that is from humans creating based on their own influences.

As pointed out in this thread, the number of recordings is effectively unlimited. So, as also pointed out in this thread, it seems that a notability bar should be set, otherwise, I fear the wheat could get lost in the chaff, so to speak.

I think the nuance here, repeating a previous point, is the machine synthesizes a new work based on existing works, in much the same way that humans perform the act of creating work. And many do so differently every time they’re prompted to. This makes the AI-as-instrument argument far harder for me to swallow.

I think this is a very good point. Somehow the level of effort an “artist” exerts (be it prompter or AI) should at least be an input into its noteworthiness or valuation.

Except synths don’t “create” music, and AI arguably does.


And as though on cue, an article reported by Reuters about ChatGPT books generated and sold on Amazon, so also a #bookbrainz concern:


This thread talks about the implications of AI content but it is not clear that this has been resolved.

Frustratingly, it looks like editors went out and created artist entries for certain AI computer programs. I do not think these are artists, as they are computer programs. They do have sleek marketing which implies that they are, or even better than humans, and we should be careful not to swallow it.

Here is one for Midjourney: Midjourney - MusicBrainz

Given this thread seems to have become a philosophical discussion, but there is no software artist type (or resolution of whether there should be), this strikes me as slightly unhelpful. To my mind, in this context, we should wait for consensus.


replying in a new, more MusicBrainz-focused thread

I created the Midjourney artists because that’s what the credit on the release says, and I want to store the data/relationship.

I don’t see what the philosophical debate here has to do with it - the relationship can always be changed if we decide to make some sort of ‘instrument’ out of these bots instead, or whatever path we go down.

If you have specific questions that could lead to a consensus of some type on how to credit the relationship today (unlike this thread), then I invite you to ask those specific questions/open those tickets. UltimateRiff’s thread has handily popped up!


Some have already created artists for other software like vocaloids. 初音ミク - MusicBrainz and IA - MusicBrainz

Those are both of type “character” and seem pretty non-controversial to me – I don’t see much difference between these artists and other ones like Bart Simpson or Alvin. I think that the key point (at least in my mind) is that both 初音ミク (Hatsune Miku) and IA seem to be discrete personas that are “managed” by companies and that have official releases.

If someone created a generic “Vocaloid” artist referring to the Yamaha software and it got used it as a dumping ground for random songs containing synthesized vocals, I might feel differently. Oh wait, that’s already happened. :slight_smile:

But you could just buy the Hatsune Miku voice and use it for your own track and then release it as by Hatsune Miku. So in that sense it’s totally not like Bart Simpson or Alivin as they would sue you into oblivion if you did that.

1 Like

Thanks, I didn’t know that they also sell the synthesizer. I’ll note that they refer to it as an “instrument” in multiple places on that page, so I don’t think it makes sense to me for unsanctioned third-party songs or releases that use the synthesizer to be credited to the fictional character.

In the news: FT: Spotify ejects thousands of AI-made songs in purge of fake streams (non-paywalled archive)

My summary: “Boomy” is a startup that lets users generate tracks from text descriptions and then submit them to streaming platforms to get royalties. Spotify recently deleted “tens of thousands” of these songs due to bogus streaming that had apparently been reported by Universal. Boomy resumed submitting tracks last weekend.

Some choice quotes that made me go :grimacing: :

California-based Boomy says its users have created more than 14mn songs.

100,000 new tracks are being added each day [to Spotify, from all sources].


So your concern is the database become petabytes larger and not having adequate funding to host it all? I mean there’s a lot of music we don’t like but if a single soul enjoys it, who are we to judge? This is not my house so I don’t set the rules but I’m sorry to see there’s a degree of prejudice before including any recorded entity in the various databases being built here. If somebody or something else is willing to input it without asking anything of you, ie. your time or your money, I’m not sure what the worry is, but it’s an amazing time to be alive, even if we’re all going to be made redundant, I’d rather live now than when those wax cylinders were the only format. I personally feel like we should all dive into all the tools within our reach to be more effective, to bring about more happiness and to unite humanity but it might be a bumpy ride if we can’t choose utopia over dystopia, liberty over authority. Good luck everyone!

Edit: A further thought -
My ghost writer has told me that the voice model that became famous as the sound of Stephen Hawking’s “robot-sounding” voice is called the “Perfect Paul” voice. It was developed by a company called Acapela Group in collaboration with Hawking himself. The voice was created by taking recordings of Hawking’s natural speaking voice and using text-to-speech synthesis to generate his iconic robotic voice. The Perfect Paul voice became closely associated with Stephen Hawking and played a significant role in enabling him to communicate despite his physical limitations.

Given that there are clearly people who are critical of using too much technology to be considered genuine, or not fraud in some way, at least cheating to some degree, then why don’t we consider all interviews with Professor Stephen Hawking after he was no longer able to “manually” speak as being interviews with Perfect Paul?
Artificial Intelligence models at the time of writing at least are programmed by humans, and so are the prompts originally.
I’m trying to convey what I’m getting at and maybe not doing a good job here. I’ll try to explain better after anybody else fancies chiming in and I’m better caffeinated because I will respect people’s wishes not to summarise what could be a long and rambling headache using GPT! Best wishes to all.

Further edit:
I believe that this video is on-topic but the title might not seem so.

I’m not worried personally about the DB becoming huge because, well, people will only add stuff they care about. As such, it’s likely the computer-generated music we get is only a minuscule amount of those 14 million songs - the ones people actually like enough to listen to. I imagine writing a good song with LLMs for example is not easy, and I have nothing against crediting the person using the LLM and requesting a ton of changes until it sounds right as the artist. As such, I think we agree on the general idea of “LLM as instrument, prompter as artist”. It will get trickier when it’s “computer-generated copy of the voice of some artist, credited as that artist”, although I guess that’ll get regulated sooner rather than later.

I think given the choice quotes, @derat’s main problem might be a thing I also worry about - LLMs make it a lot easier to make zero-effort music for spammy purposes. This is something we’re already fighting against in MusicBrainz - people who are not music-related at all otherwise will make a computer-generated music track or two (in seconds, because it doesn’t have to be good for it to work), then add themselves to MusicBrainz because they have music now, with all their links for SEO (which is the actual purpose of their MusicBrainz page for them anyway). More widely, how easy it is to make zero-effort tracks will probably make it harder to discover good music, LLM-aided or not - but I guess that just makes ListenBrainz’s recommendation tools more important than ever.


I don’t know if we’re 100% in agreement here - if prompting requires some kind of effort to have AI produce a useable song, then yes, this is probably the way to go.

In the more likely case (IMO) where it takes little to no effort on the part of the prompter, where other AI like chatGPT can write the prompts, or the AI is pumping out songs without prompts, I don’t think see a way that MB will be able to wriggle out of having AI in artist credits. At least, not without a schema change.

That just sounds like [no artist] to me :smiley:


A post was merged into an existing topic: How to handle AI in MusicBrainz

@reosarevok this topic was rather meant to be general (meta).

Or as someone put it, “a philosophical discussion”, not just music :slight_smile: (Even if that’s what’s currently most talked about)

For example, recently OTW decided to hinder scraping of AO3 data for AI purposes and might remove such “AI” works if they appear too spammy:

Some recent examples of hallucination gone awry:

There might be good use cases for ChatGPT but facts are not one of its strong points.

1 Like
1 Like