AI Derivatives, Licenses, and the Shape That Listening Is Taking
I have breakfast late. The house is still in that phase where sounds seem farther away than they are. I play a track from Discovery Weekly, one of those Spotify recommendations every Monday with a clean face and dirty hands: “I know you,” it says, “and yet I surprise you.”
A soft groove starts, a drumbeat that comes in without hurry. I scroll through the feed with the same hand I use to tear bread, a double, automatic gesture. News, releases, recommendations, a bit of noise dressed as information
The word that sticks is a technical one, almost cold: derivatives.
Derivatives. Versions.
Then a passage appears regarding Spotify’s results list, and my thumb slows, as if it had heard a sour note change the kitchen air. It’s not just any news. It’s one of those phrases that seem “just business” and yet make you reflect.
The article talks about two people, two roles. The first is Gustav Soderstrom, co-CEO of Spotify. He explains how the company sees musical AI and where it sees room. The second is Alex Norstrom, also co-CEO. He reiterates that deals must be “good for artists.”
I stay a few seconds there, coffee halfway. Because when a company like Spotify talks about music AI in that context, it is rarely just a discussion about the future. It is probably beginning to rewrite what its next normal will be.
Soderstrom divides the field. On one side, music created “from scratch” with AI tools. On the other, derivatives: covers, remixes, variations generated from existing tracks. And it is clear from the start why the second category is the most delicate. Because there nothing is invented from scratch; the work is based on memory, audience, and value. You enter the catalog like you enter a stuffed warehouse.
The track in the headphones goes on and I begin to see Spotify with different eyes. Until yesterday it was mainly a space for musical research. Today, in my eyes, it is something else entirely: a place where music can become working material, and listening takes on the characteristics of an assembly line, made of quick choices and infinite variations.
It is easy to imagine the concrete scene. A button, an integrated function:
“Make me a version in the style of…”,
“Make it slower”,
“Put it in a different mood.”
The music that comes to me is no longer just a track: it is a bundle of possibilities. And here is where the central hitch lights up: here the version becomes an operation.
Every version is a repeatable operation. And when this operation repeats continuously, someone optimizes the flow and decides who earns.
If the gesture becomes replicable, industrial logic steps in: rules and tolls. At that point the key question arises: who controls the tap?
The catalog, in this story, is no longer just a cultural archive but capital. A heritage that can produce revenue in new forms. Cinema, for a long time, has already arrived there, through sequels and remakes that reactivate already known stories. The difference is that the operation can happen on a daily scale, in the smallest gesture: me having breakfast and generating a variant.
A button is needed.
Meanwhile, under the button, grows something that usually no one wants to observe because it spoils the magic: infrastructure.
Every variant requires computation. And computation means servers, storage, cooling, bandwidth, and moderation. These are technical words, but the translation is: energy, capacity, and control.
The single gesture seems small but the sum of millions of gestures every day becomes a heavy system, made of real costs and real decisions. And those who bear those costs usually put themselves in the position to collect the value that derives.
Thus a thought becomes even more concrete: the discussion about licenses is not a detail: it is the lock. The point where it is established what can be done with a track, who can transform it, under what limits, with what economic returns. It is the part where it is decided whether participation is freedom or refined consumption.
Alex Norstrom says that deals must be “good for artists.” The phrase seems almost like an institutional caress. Reading it, I try to translate it into operational and more direct questions. I wonder, therefore:
Good in what sense?
Which artists?
According to which metric?
Good when the system begins to produce variants at feed speed.
Because the feed has already educated listening. The idea that every track is a quick test, a taste, an instant decision.
If tomorrow ten versions of the same track can exist, the natural temptation will be to treat them like content: try, discard, move on. And in that gesture of discard there is a cultural transformation bigger than any debate about sound quality.
At that point listening becomes a behavior. A way of being. A rhythm.
When the second track starts, I begin to glimpse the scenarios that follow, like lit-up shop windows.
I see a market of variants.
I see huge catalogs becoming central, because they are full of recognizable material
Recognizability wins because it reduces friction and retains.
It will rain on the wet. Strong tracks generate more versions, the versions push the strong tracks, and the system finds a new form of revenue.
Then I see the word fandom change meaning. Creativity leaves measurable traces. It produces data that manipulates listening decisions on platforms.
Here a deep and subtle question arises: can this variant economy generate a music scene?
Historically, scenes are born from three elements difficult to compress into an interface: proximity, risk, time. Shared places, forming languages, repeated mistakes that become style.
A variant platform instead tends to produce another type of novelty.
There is perceptual newness. An unexpected combination, a different color, something that makes you say “oh.” And there is cultural newness. That which creates lexicon, communities, influence. That which resists.
The latter does not arise only from combining sounds. It arises from context, risk, time, friction. It arises from the fact that someone persists, fails, returns, builds a scene, a shared language. When the variant becomes easy and continuous, the risk is that the new remains often at the first level: brief surprise, quick spark, then move on.
Here I remember my kitchen, my breakfast, my hand on the feed. The matter does not only concern industry. It concerns the way I—and many others—are trained to listen.
The infinite variant becomes a form of comfort. A small creative gratification, a way to accelerate time and speed everything up.
Like a bead kit: you choose which palettes to compose and combine at your pleasure. The perimeter is already drawn. Freedom coincides with parameters, and parameters coincide with governance. Someone decides what is available, what is recommended, what is pushed. Someone decides which transformations are lawful and remunerated.
At this point the discussion ceases to be polarized: pro or against AI. It becomes rather a discussion about the hitches of the proposal. A proposal based on the variant economy: a model that lives on volume and tends to reward what is already strong, what adapts to recognizability and favors permanence.
The only adult answer is not morality: it is to demand a few practical and verifiable conditions.
Readable traceability. Origin, transformation, license in the clear. A threshold between generating and publishing. Directorial rights for artists, granular and revocable.
Spotify says the rights framework is missing. The framework decides whether this season will see enduring works or variants with collective participation similar to a subscription.
Monday morning returns. Coffee, light, headphones.
Every time you request a version, you move a grain. Millions of grains become a dune. And a dune changes the landscape even when no one is looking.
Articolo scritto da Giampiero Stramaccia, editato da Eleonora Poli.
Ufficio Stampa Dancity – press@dancityfestival.com
Dona con PayPal
Il nostro progetto è indipendente e senza fini di lucro.
Se ti piace quello che facciamo, puoi aiutarci con una piccola donazione.
Questo articolo è stato prodotto con il supporto di strumenti di intelligenza artificiale, utilizzati per l’organizzazione dei contenuti e l’ottimizzazione testuale. Le fonti, le idee e i materiali provengono dall’archivio e dall’attività dell’Associazione Dancity.
