AI Generated / Assisted Music

kolaval said:
Would you listen to an AI song if it was good?
Also if you do not know the difference, does it matter?
Click to expand...
I've used Suno before for fun, but I don't see it as real music and more it's just a curiosity as a technology.

Real music takes creativity which prompting is not a replacement for (nevermind the argument of using real artists music for training without their consent).

If I'm trying to discover new music I want to be able to have the opportunity to see that act live. Real humans convey their own emotions in their music and that's such a huge part of the listening experience, of which AI music has zero of. It's just empty.
 
DrJohnZoidberg said:
I've used Suno before for fun, but I don't see it as real music and more it's just a curiosity as a technology.

Real music takes creativity which prompting is not a replacement for (nevermind the argument of using real artists music for training without their consent).

If I'm trying to discover new music I want to be able to have the opportunity to see that act live. Real humans convey their own emotions in their music and that's such a huge part of the listening experience, of which AI music has zero of. It's just empty.
Click to expand...
I think you're conflating the tool with the person using it.

If someone types "make me a sad rock song" into Suno, takes the first result and uploads it, then sure, there's very little creative input from that person.

But that's not the only way people use these tools. I write and rewrite lyrics, decide what the song is actually about, work on structure, experiment with genres and instrumentation, generate multiple versions, reject most of them, edit and arrange what works, and sometimes take the result into a DAW and work on it further. Those are creative decisions. The fact that I didn't personally play the guitar doesn't somehow make those decisions cease to exist.

And "AI can't feel emotions" isn't really the killer argument people think it is. My camera doesn't feel anything either. Photoshop doesn't understand why a photograph matters to me. A synthesiser doesn't experience melancholy. They're tools. The emotional intent comes from the human deciding what they want to express and making choices about the result.

You can absolutely prefer music written and performed entirely by humans. You can value knowing the artist's story, seeing them live, watching musicians perform and having that human connection. I do understand that, and AI music can't reproduce all of those things.

But that's an argument for why you value traditional musicianship. It isn't evidence that music made with AI assistance is inherently "empty".

The training-data/consent issue is also a legitimate discussion, but it's a separate ethical question from whether somebody can use the technology creatively.

AI makes it possible to create music with almost no effort. It doesn't follow that everyone using AI puts almost no effort or creativity into making music.
 
OnlyOneKenobi said:
The training-data/consent issue is also a legitimate discussion, but it's a separate ethical question from whether somebody can use the technology creatively.
Click to expand...
Before any other arguments even begin this an absolute non-starter.

It feels utterly gross listening to something in a serious capacity knowing that it's generated by a tech company profiting off existing works without a care in the world. When Suno and its competitors started up I thought it was fun excercise that would be sued out of existence in a matter of months, but it's apparently true that if you have enough money you can just do what you want.
 
DrJohnZoidberg said:
Before any other arguments even begin this an absolute non-starter.

It feels utterly gross listening to something in a serious capacity knowing that it's generated by a tech company profiting off existing works without a care in the world. When Suno and its competitors started up I thought it was fun excercise that would be sued out of existence in a matter of months, but it's apparently true that if you have enough money you can just do what you want.
Click to expand...
I genuinely don't understand the need to repeatedly visit topics about something you clearly dislike just to tell everyone how much you dislike it. But anyway...

Your objection is rapidly becoming obsolete even on its own terms.

Suno has now partnered with Warner Music Group, BMG and Believe/TuneCore, with its newer generation of models being developed through music-industry partnerships. Warner's agreement specifically provides for licensed models, compensation and opt-in protections for artists.

So if your objection is genuinely that you cannot ethically listen to AI music because artists or rights holders didn't consent, what happens when they do consent, license their material and get paid?

Does the resulting music suddenly stop being "empty"?

Because if not, the training data argument was never actually the reason you considered it empty. It's simply another objection to AI music.

I'm also not particularly persuaded by the idea that learning from existing creative work is inherently unethical. Human creativity doesn't emerge in a vacuum either. Musicians spend their lives listening to other musicians. Writers read other writers. Photographers study other photographers. Every creative person absorbs an enormous body of existing work and develops their own ideas from it.

Obviously an AI model does this technically and at a scale completely unlike a human brain, and copyright law may reasonably draw distinctions between the two. But "it learned from copyrighted material" isn't by itself enough for me to conclude that everything subsequently produced with it is morally contaminated.

More importantly, the industry itself is already moving towards licensing rather than making the technology disappear. Warner went from suing Suno to partnering with it. That's a fairly significant development.

You're entirely entitled to dislike AI-generated music. But we've now gone from "it's not real music", to "there's no human creativity", to "there's no emotion", and now to "the training is unethical".

If licensed models remove the ethical objection and you still consider the result inherently empty, then perhaps the objection was never really about the training data in the first place.

Perhaps you simply don't like AI music. Which is fine. 🙂
 
OnlyOneKenobi said:
I genuinely don't understand the need to repeatedly visit topics about something you clearly dislike just to tell everyone how much you dislike it. But anyway...

Your objection is rapidly becoming obsolete even on its own terms.

Suno has now partnered with Warner Music Group, BMG and Believe/TuneCore, with its newer generation of models being developed through music-industry partnerships. Warner's agreement specifically provides for licensed models, compensation and opt-in protections for artists.

So if your objection is genuinely that you cannot ethically listen to AI music because artists or rights holders didn't consent, what happens when they do consent, license their material and get paid?

Does the resulting music suddenly stop being "empty"?

Because if not, the training data argument was never actually the reason you considered it empty. It's simply another objection to AI music.

I'm also not particularly persuaded by the idea that learning from existing creative work is inherently unethical. Human creativity doesn't emerge in a vacuum either. Musicians spend their lives listening to other musicians. Writers read other writers. Photographers study other photographers. Every creative person absorbs an enormous body of existing work and develops their own ideas from it.

Obviously an AI model does this technically and at a scale completely unlike a human brain, and copyright law may reasonably draw distinctions between the two. But "it learned from copyrighted material" isn't by itself enough for me to conclude that everything subsequently produced with it is morally contaminated.

More importantly, the industry itself is already moving towards licensing rather than making the technology disappear. Warner went from suing Suno to partnering with it. That's a fairly significant development.

You're entirely entitled to dislike AI-generated music. But we've now gone from "it's not real music", to "there's no human creativity", to "there's no emotion", and now to "the training is unethical".

If licensed models remove the ethical objection and you still consider the result inherently empty, then perhaps the objection was never really about the training data in the first place.

Perhaps you simply don't like AI music. Which is fine. 🙂
Click to expand...

lol people against it is starting to sound like vegans.....they have to tell you about it, or just a general shitstirrers, either way I pay it no attention.

It is usually the same people complaining about it, have absolutely limited or no experience whatsoever. Trying to explain to people there is more to prompting than simply outputting text and some more advance prompting requiring music knowledge.
 
OnlyOneKenobi said:
So if your objection is genuinely that you cannot ethically listen to AI music because artists or rights holders didn't consent, what happens when they do consent, license their material and get paid?

Does the resulting music suddenly stop being "empty"?
Click to expand...
It continues to be empty no matter what.

Here are some analogies to help you understand:

Is climbing Mount Everest an achievement? - Yes
Is getting a helicopter ride to to the top of Mount Everest as a passenger and achievement? - No

Is being an astronaut an achievement? - Yes
Is Katy Perry an astronaut? - No

Is running the Comrades an achievement - Yes
Is completing the Comrades route in a car an achievement - you can hopefully see the pattern now... - No

Every single phone on this planet has enough computing power to destroy every human at chess but in reality nobody consumes content or creates a culture around watching computers play chess against each other. It doesn't exist. However there are literally millions of people who idolize chess grandmasters and the community of chess. Why? Because the achievement of a human is not empty and invokes an emotional and human response. The humanity of it is what gives it value.

AI music might exist but the market for it is extremely small and ultimately pointless. A waste of scare resources to create it.

If you had a daughter and she spent 10 years learning piano would you as a parent be ok with her competing against a piano playing machine in the music eisteddfod? I hope not. And it is for this exact same reason why it is wrong for AI music to be generated based off the hard work of humans. Whether permission is given or not. That doesn't change how empty and meaningless it is.
 
AlmightyBender said:
It continues to be empty no matter what.

Here are some analogies to help you understand:

Is climbing Mount Everest an achievement? - Yes
Is getting a helicopter ride to to the top of Mount Everest as a passenger and achievement? - No

Is being an astronaut an achievement? - Yes
Is Katy Perry an astronaut? - No

Is running the Comrades an achievement - Yes
Is completing the Comrades route in a car an achievement - you can hopefully see the pattern now... - No

Every single phone on this planet has enough computing power to destroy every human at chess but in reality nobody consumes content or creates a culture around watching computers play chess against each other. It doesn't exist. However there are literally millions of people who idolize chess grandmasters and the community of chess. Why? Because the achievement of a human is not empty and invokes an emotional and human response. The humanity of it is what gives it value.

AI music might exist but the market for it is extremely small and ultimately pointless. A waste of scare resources to create it.

If you had a daughter and she spent 10 years learning piano would you as a parent be ok with her competing against a piano playing machine in the music eisteddfod? I hope not. And it is for this exact same reason why it is wrong for AI music to be generated based off the hard work of humans. Whether permission is given or not. That doesn't change how empty and meaningless it is.
Click to expand...
I think the problem with this argument is that you're treating AI music creation as though it is the musical equivalent of asking an image generator to "draw me a sad woman."

Yes, music prompting can be that simple. You can type "make me a sad song" into an AI music system and it will happily produce something reasonably generic. But that's not really where serious AI music creation ends.

The creative direction can get considerably more involved because you're directing the actual musical construction. Understanding things like BPM, time signatures, chord progression, chord voicing, pitch movement, vocal processing, echoes, delays, sidechain compression, rhythmic patterns, instrumentation, arrangement, tension and release, drops, transitions and how different sounds interact all materially changes the result.

For example, I have a 3-step track that uses bagpipes as part of the instrumentation. That wasn't the AI randomly deciding that bagpipes would be hilarious. The choice was deliberate because the instrument gives the track a particular hook and character. You need to understand enough about music to know why an unusual instrument might work and where to put it.

suno.com

Letsatsi La Ka

Listen and make your own on Suno.
suno.com suno.com

With an image generator, you can describe what you want the finished image to look like in considerable detail. Once you've specified the visual concept, much of the creative process is still inside the generator.

AI music is increasingly different. You're often using the AI to fill in musical blanks while retaining a much larger amount of creative direction over the construction of the piece. You're not simply specifying the finished product. You're influencing how the thing is musically built.

And the technology has moved considerably beyond "type a prompt and receive a song."

You can now bring your own samples and loops into the process, incorporate your own musical material, and in some systems use your own voice. Suno in particular has developed far beyond being a pure text-to-song generator. A lot of the workflow is starting to resemble elements of a DAW, with generation becoming one component of a broader creative process rather than the entire process.

That's also why the "my daughter spent ten years learning piano, therefore AI music is meaningless" analogy doesn't really work.

If the competition is specifically testing human piano performance, then obviously a piano-playing machine shouldn't enter. The machine isn't demonstrating the skill the competition exists to measure.

But that's completely different from saying that somebody using a new creative tool has produced something "empty."

Nobody claims that somebody who uses a DAW has personally played every instrument in their track. Nobody claims that electronic musicians aren't musicians because they use synthesizers, samples, effects, automation, quantization or software instruments.

And this is where the history of electronic music becomes rather inconvenient for the argument.

A huge amount of modern music production already relies on technology doing things automatically that previously required considerably more manual work. EDM didn't suddenly become "not music" because people started sequencing music inside DAWs instead of recording an entire band live.

Even genres such as amapiano and gqom didn't emerge from a world where every producer was a conservatory-trained professional musician sitting behind an acoustic piano. A substantial amount of electronic music has historically been created in bedrooms, often with inexpensive or even pirated copies of software such as FL Studio, using samples, loops, synthesizers, sequencing and automation.

DAWs themselves have steadily moved from being tools associated primarily with EDM and electronic production into general-purpose music production environments used across practically every genre.

So yes, there is absolutely a legitimate discussion to have about AI training data, copyright, licensing, attribution, compensation and what constitutes authorship.

But "AI music is meaningless because the machine did the difficult part" is not really an argument about AI music. It's an argument based on a particular definition of what counts as musical achievement.

And if that definition is "the more physical effort and technical skill required to produce the final sound, the more valuable the music must be," then you're going to have to throw a surprisingly large chunk of modern music production under the bus too.

The biggest problem here is that you're confidently critiquing a creative system that, from your description, you appear to have very little experience actually using.

AI music isn't simply "press button, receive song."

At least, it doesn't have to be.


If you can't grasp the concept perhaps keep your trap shut, trust me, it wouldn't end well for you. If I have to start getting technical about the aspects of song creation, good fcking luck brother, you will be in for a rough day.
 
AlmightyBender said:
It continues to be empty no matter what.

Here are some analogies to help you understand:

Is climbing Mount Everest an achievement? - Yes
Is getting a helicopter ride to to the top of Mount Everest as a passenger and achievement? - No

Is being an astronaut an achievement? - Yes
Is Katy Perry an astronaut? - No

Is running the Comrades an achievement - Yes
Is completing the Comrades route in a car an achievement - you can hopefully see the pattern now... - No

Every single phone on this planet has enough computing power to destroy every human at chess but in reality nobody consumes content or creates a culture around watching computers play chess against each other. It doesn't exist. However there are literally millions of people who idolize chess grandmasters and the community of chess. Why? Because the achievement of a human is not empty and invokes an emotional and human response. The humanity of it is what gives it value.

AI music might exist but the market for it is extremely small and ultimately pointless. A waste of scare resources to create it.

If you had a daughter and she spent 10 years learning piano would you as a parent be ok with her competing against a piano playing machine in the music eisteddfod? I hope not. And it is for this exact same reason why it is wrong for AI music to be generated based off the hard work of humans. Whether permission is given or not. That doesn't change how empty and meaningless it is.
Click to expand...
I understand the analogies perfectly well. I just don't think they establish what you think they establish.

You're comparing achievement in performing a physical or technical task with the creative value of an output. Those aren't the same thing. Nobody is claiming that generating music with Suno is equivalent to spending ten years mastering the piano, any more than taking a photograph makes me a painter because I didn't spend three days reproducing the scene with oils.

Your Everest analogy would work if I generated a piano performance and then claimed that I had personally mastered the piano. I haven't, and I wouldn't. But we're discussing whether something created using AI can have creative or emotional value. That's a completely different question.

The daughter/eisteddfod example makes exactly the same category error. I wouldn't enter an AI-generated piano performance into a competition intended to measure human piano performance. Obviously. Just as I wouldn't enter a photograph into a portrait-painting competition and argue that my camera painted it. That tells us absolutely nothing about whether photography has artistic value.

Your chess example is particularly strange because computer chess very much does exist. Humans play against engines, analyse engine games, hold computer chess championships and use engines extensively to develop human chess. What people generally don't do is pretend Stockfish is a human grandmaster. Nobody needs to pretend Suno is a human musician either.

Your position appears to be that the value of music necessarily comes from the human achievement involved in physically learning to perform or compose it. That's a perfectly coherent personal philosophy of music. But it's not an objective definition of artistic or emotional value.

It also leads to some very awkward boundaries. How much automation is allowed before the work becomes empty? Drum machines? Sequencers? Arpeggiators? Sampling? Pitch correction? Algorithmic composition? Generative synthesis? DAW quantisation? Session musicians playing something somebody else wrote? A producer constructing a track from performances and samples they didn't personally play?

There's no obvious point along that continuum where human creativity magically disappears. For me, the interesting question isn't "how difficult was the tool to operate?" It's what the human actually contributed.

Someone typing "make rock song" and accepting the first generation has contributed very little. Someone writing the lyrics, defining the concept, directing the arrangement and sound, generating and rejecting material, editing it, restructuring it, combining generations and finishing the production in a DAW has made considerably more creative decisions.

Neither person has thereby become a pianist, guitarist or vocalist. They haven't earned those achievements and shouldn't claim them. But "you didn't acquire the traditional skills required to produce every component yourself" does not logically lead to "therefore the thing you made is empty and meaningless".

And no analogy about driving the Comrades route changes that.
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X