AI + Music --> What will the future look like?

[VariKusBrainZ]

Or we could just turn it off

[annode]

Pages of predictions. But can we really predict something that didn't exist before? No one could predict how new technologies will create music of a kind that has never before been thought of. I see the same with Ai. How it gets used for such an innovation can not be know until it happens. Whatever it will be will be a total surprise.

[vurt]

[Grizzellda]

By the way vurt, as I said...(yeah I'm an asshole)

You got alot of posts thar... I tried to count but alas... I could not count that high...

Just clowning with you... also, inhaling "buds"...

[enroe]

Pages of predictions. But can we really predict something that didn't exist before? No one could predict how new technologies will create music of a kind that has never before been thought of. I see the same with Ai. How it gets used for such an innovation can not be know until it happens. Whatever it will be will be a total surprise.

Well, if the thread title is "What will the future look like?"
or "Will Ai be the next big step in music?", then it's precisely
about looking into the future and making predictions.

One has to dare to make predictions. Almost all political
decisions are based on predictions and models of the future.
So the only question is:

"How reliable are such predictions?"

Ultimately, predictions are always subject to uncertainty,
you're absolutely right about that. However, to prevent the worst,
one must definitely strive for predictions.

If an asteroid were hurtling towards Earth, it would be vital for
survival to be able to predict whether it would actually hit, what
its mass would be, etc.

And it's a bit like that with AI too: We can already see in many
areas of the economy and the media landscape how AI works,
what effects it has. And we can extrapolate from this what these
effects will probably look like in a few years.

Similarly, we can already see how SUNO, UDIO, Loudly, Canva,
etc., automatically generate songs. Initially, almost all musicians
who play an instrument themselves have an aversion to it, but
then individual AI parts are adapted for their own songs (examples
can even be found in this forum!). Eventually, however, entire
songs are generated solely via text prompt.

In just a few years, these songs generated via text prompt will
be indistinguishable from songs organically recorded in the
studio – and the general public, the consumers, will no longer
care whether good-sounding music is AI-generated or not.

If you also consider the laws of our economy, then it is actually
no longer an unlikely or even daring assumption to say: "AI will
take over all commercial song composition. Humans as song
composers will no longer play a role."

[Bunny_boy]

But can we really predict something that didn't exist before?

Ar you talking about genres of music? Humans do that quite frequently

[Zeisner]

In just a few years, these songs generated via text prompt will be indistinguishable from songs organically recorded in the studio

For this to happen, all of the following is required:

-> A lot more data centers. Right now it doesn't look so good for AI companies when it comes to attracting new investors because of the still missing revenue.

-> A lot of new and unprocessed audio data, primarily to reduce artifacts. There is only one way to get this for free (or rather cheap): To turn all the major OSes and DAWs into data miners, sending everything the user ever makes to the data centers. Very difficult to realize without pushing too many people towards Linux or into offline mode.

-> Hiring a lot of musicians and audio engineers for supervised learning and additional model finetuning. This can be clearly ruled out because nobody is going to spend a single cent on that.

The good (or bad) news is: It's not even necessary to raise quality that much. Most people can't hear the differences anyway. No artifacts, no weird lifeless performance. Most people can't even differentiate between a Minimoog and an electric guitar. Or a real drum kit and a 808.

[Bunny_boy]

In just a few years, these songs generated via text prompt will be indistinguishable from songs organically recorded in the studio

For this to happen, all of the following is required:

-> A lot more data centers. Right now it doesn't look so good for AI companies when it comes to attracting new investors because of the still missing revenue.

-> A lot of new and unprocessed audio data, primarily to reduce artifacts. There is only one way to get this for free (or rather cheap): To turn all the major OSes and DAWs into data miners, sending everything the user ever makes to the data centers. Very difficult to realize without pushing too many people towards Linux or into offline mode.

-> Hiring a lot of musicians and audio engineers for supervised learning and additional model finetuning. This can be clearly ruled out because nobody is going to spend a single cent on that.

The good (or bad) news is: It's not even necessary to raise quality that much. Most people can't hear the differences anyway. No artifacts, no weird lifeless performance. Most people can't even differentiate between a Minimoog and an electric guitar. Or a real drum kit and a 808.

Data centre resource use will reach a peak due to them becoming unsustainable, or there will be huge huge problems. Wonder if the Liddites will make an appearance again? I think your point about them hiring Audi pros is akin to that - nobody would want to be involved with that as would be helping destroy the music industry.

Amazon is feeling something related to the AI concerns at the moment.
https://www.bbc.co.uk/news/articles/c150e144we3o

[Grizzellda]

And as for these "data centers"...

They may be sustainable for now, but as time goes on?

[Zeisner]

Data centre resource use will reach a peak due to them becoming unsustainable, or there will be huge huge problems.

There are already huge problems with residents having bad water or experiencing power shortages.

I think your point about them hiring Audi pros is akin to that - nobody would want to be involved with that as would be helping destroy the music industry.

This and the fact that they would have to work for free and with some of the biggest idiots in the world: Software engineers.

[DCrown]

Tomorrow will look like today, the only difference 10 Mio new uploads every day instead of about 150.000 (now) new audio/sound/noise/some call it song uploads every day

[enroe]

Tomorrow will look like today, the only difference 10 Mio new uploads every day instead of about 150.000 (now) new audio/sound/noise/some call it song uploads every day.

Yes, at Spotify.

And 99% of them are created by AI-text-prompts.

[DCrown]

I would like to know how many songs were released in 1970s and 1980s in one year, I doubt it was even close to 150.000.

[DCrown]

Google tells me about the 1980s:

Billboard Hot 100: Over the entire decade, approximately 4,172 different songs appeared on the Billboard Hot 100 chart. This averages to about 417 charting songs per year.

Of course more songs were produced than 417 per year.

Vinyl was dominant in the early 80s, with 322 million units sold in 1980, before falling to 34 million by 1989 as CDs and cassettes took over.

Well, times are changing.
The day I will start using AI, I will easily produce more than 417 songs per year.
Let's say 15-20 new songs every day, it will take me less than an hour every day.

[ghettosynth]

... why you are doing music and so forth, so the listeners have a connection to a human and not an algorithm.
...
I think one example is the connection to art and artists -- for example last night I read stories about Miro the painter. I would not read real life stories about Claude's LLM algorithm.

Have you heard of Bastien Soryn?

Bastien Soryn is a composer, sound archivist, and former systems engineer whose work centers on the preservation of human presence in an increasingly automated cultural landscape. Born in a declining shipyard city along a cold northern coast, he grew up surrounded by mechanical noise: turbines, freight cranes, the metallic percussion of industry. His mother restored antique pianos in a small workshop behind their apartment, and from an early age he absorbed two parallel languages, the precision of machines and the fragility of human touch.

He studied electrical engineering with a focus on signal processing, eventually contributing to the development of large scale audio classification systems for media platforms. His role involved designing architectures that could analyze millions of tracks for mood, tempo, and engagement metrics. The work was technically rigorous and commercially successful. It was also disquieting. Bastien began to notice how patterns flattened nuance, how statistical averages erased the hesitations and imperfections that made a voice recognizably human.

In his early thirties, after the death of his younger sister, he withdrew from corporate research and enrolled in a conservatory program in composition and ethnomusicology. He spent several years traveling through remote communities, documenting regional songs that had never been recorded. He became particularly interested in the microvariations of live performance: the breath taken half a beat too soon, the subtle drift in pitch when emotion overrides training. To him, these were not errors. They were signatures of presence.

Bastien’s music is sparse, deliberate, and structurally restrained. He avoids algorithmic generation tools in his own compositions, not out of hostility toward technology, but from a conviction that authorship carries moral weight. He has written essays arguing that listeners seek more than sonic arrangement. They seek evidence of another consciousness, however distant. In interviews, he states that art is a trace of vulnerability, and that when creation is delegated entirely to autonomous systems, that trace becomes ambiguous.

Today, Bastien Soryn divides his time between composing chamber works, lecturing on digital ethics in creative industries, and maintaining a private archive of endangered recordings. He does not frame the tension between human and machine as a battle. He frames it as a question of responsibility. For him, making music is an act of accountability. It is a declaration that someone chose each note and accepted the consequences of that choice.