A singer steps into the light. We see a face, hear a voice and assume they belong together. AI has made that certainty harder to hold—but the question began long before AI.
Something strange is happening to music.
You can pull out your phone, type a few words or hum a melody, and get a complete song back. It can have lyrics, drums, guitars, strings, backing vocals and a huge chorus. It can sound ready for radio, even if somebody made it while sitting on a bus.
But the finished sound no longer tells us much about how it was made.
Is that a real voice? Did the person write the words? Did they create the melody? Did they play anything? Did they feel what the song is saying? Is there even a person behind it?
These questions feel completely new.
Before we decide whether they are, we need to go back to 1978, back to “Ra-ra-Rasputin.
Boney M
In 1978, Boney M released “Rasputin.” It became a huge hit then, and nearly fifty years later its hook is still moving through clubs, memes, dance videos and new generations of listeners.
Most people who watched Boney M saw Bobby Farrell at the centre. He was all movement and high energy. They saw his body, heard a deep male voice and naturally joined the two together.
The man they saw must have been the man they heard.
He was not.
The recorded male voice belonged to Frank Farian, the German producer behind Boney M. Farian also co-wrote and produced “Rasputin.” Farrell gave that voice a face, a body and an energy it did not have in the studio.
To the audience, the face, the voice, the movement and the song arrived as one complete thing. In more formal language, they looked like an integrated object: many separate parts made to feel as though they naturally belonged together.
But they had been assembled.
This raises an uncomfortable question. Would “Rasputin” have pulled people in the same way if Farian himself had stood at the front and performed it?
The hook might still have been catchy. The record might still have sounded good. But it would not have felt like the same thing. The song needed a body that matched its energy. Farrell made the sound visible.
The connection between the body and the voice was not real. But it completed the experience, so the audience accepted it as real. Perhaps, at some level, we wanted it to be real.
We did not only want the song.
We wanted someone to be the song.
That brings us close to the problem we now have with AI music. AI can produce a strong melody, a polished arrangement and a convincing voice. But once we hear that a song was made by AI, we often pull back. There is no obvious person to place inside it.
Boney M gave the machinery a human face.
AI arrives with the machinery showing.
To understand why that difference matters, we need to stay with Frank Farian. About a decade after “Rasputin,” he used the same basic structure again. This time, the audience did not forgive it.
Milli Vanilli
Frank Farian created Milli Vanilli with Rob Pilatus and Fab Morvan at the front. They supplied the faces, bodies, dancing and personalities. Other singers supplied the voices on the records.
Again, the faces were not the voices.
At first, the arrangement worked. The songs became hits, the performers became stars, and the audience experienced everything as one complete package.
Then the package broke open.
When the truth came out, it became a global scandal. Milli Vanilli lost its Grammy, its career collapsed, and the two visible performers became the punchline.
But what had really changed?
The music industry had not suddenly started using teams. It had always used teams. Farian had already separated a male studio voice from the male body audiences watched in Boney M.
What changed was that the audience saw the split.
With Boney M, the parts stayed joined in the public imagination. People formed an attachment to the complete picture, and “Rasputin” survived. With Milli Vanilli, the picture broke apart in public, and people felt deceived.
That contrast helps explain something about AI music.
People still dance to “Rasputin” after learning that Farrell was not the recorded male voice. Yet many people will reject an AI song the moment they see the label, even if they enjoyed the sound seconds earlier.
So our response is not based on sound alone.
It also depends on what we believe the sound came from.
With Boney M, the machinery stayed behind the curtain. With Milli Vanilli, the curtain tore. With AI, there may be no curtain in the first place.
The voice may no longer have a face at all.
Milli Vanilli was treated as a bizarre exception. But once we stop looking only at the scandal, the same separation appears everywhere. Most of the time, it is simply ordinary enough that we no longer notice it.
The Pattern Never Stopped
A single artist can stand at the front of a song while writers, producers, musicians, engineers and marketers build the work around them. The audience hears one person. The credits describe a system.
Rap makes the tension especially clear because the words often sound like personal testimony: this is what I saw, what I survived and who I am. But a track can still contain several writers, producers, samples and featured artists. We hear one person telling us the truth of their life, even when many people helped construct what we hear.
The distance between the performer and the source becomes even clearer when the performance is not only presenting a voice, but an entire identity.
That is where t.A.T.u. enters the story.
In the early 2000s, “All the Things She Said” became a global hit. It presented two teenage girls in love, willing to stand against a world that rejected them.
At the time, neither performer identified as queer. They had not written the song, and the public story around them had been constructed by their producer.
Yet the song meant something real to many listeners. Some heard it and thought:
That is what I feel. I just did not know how to say it.
The image was manufactured. The listener’s feeling was not.
This takes the argument beyond fake singing or hidden songwriters. People do not only receive a sound. They receive a claimed human source for that sound. They believe somebody felt it, lived it and found a way to express it.
When that source turns out to be constructed, the listener’s experience does not become false. But the relationship was built on a mismatch.
The performers were presenting an identity that was not theirs at the time.
The listeners may have been living it for real.
Once we see that, the present begins to look different. AI did not invent the distance between the work and the person standing in front of it. It changed how much machinery was needed to create the work—and made the distance much harder to hide.
The Team Moves Inside the Phone
AI can help write lyrics, build melodies, create instruments, reshape arrangements, tune voices, fix timing, mix tracks and produce endless versions of an idea.
A person does not need a studio.
They do not even need a laptop.
The jobs themselves are not new. Teams have been doing them for decades.
What is new is that much of the team can now sit inside one tool, and that tool can sit inside a pocket.
AI has not removed the production system. It has compressed parts of that system into software.
Work that once needed expensive equipment, skilled workers, industry contacts and a large budget can now be directed by one person on a phone. The industry’s great advantage was not only talent. It controlled the machinery that could turn a rough idea—or sometimes simply a marketable face—into a polished product.
Now that machinery is becoming available to everyone.
That is the real source of the panic.
The industry sold us the output of a large system through a single human face. AI can now produce parts of the same output without needing the large system, and often without giving us the face.
The old process hid its fragmentation.
The new process makes it obvious.
At this point, the argument begins to move beyond music. The same confusion appears wherever AI starts doing work that used to be done by a team. Writing gives us perhaps the clearest
“AI-Generated” and “Editor-Generated”
Editors have always changed books.
They improve sentences, cut weak paragraphs, fix transitions, move chapters, point out missing ideas and sometimes reshape the whole structure.
We still call the result the author’s book.
We never pick it up and say:
This book was editor-generated.
But when AI performs some of the same tasks, we may call the whole book AI-generated. The work can be similar. Only the way the help arrived has changed.
The distinction becomes even harder to defend when professional editors use AI themselves. You can send a manuscript to an editor, have that editor run parts of it through AI, and receive the work back under the trusted name of human editing.
If any use of AI makes a work AI-generated, then eventually almost every book, song, film and piece of software will fall into that category. The label will become too broad to tell us anything useful.
At that point, the question has to change.
Not simply:
Was AI used?
But:
What did the human being bring?
What did they notice? What did they choose? What did they reject? What holds the work together? Can they explain it, defend it and take responsibility for it?
Those questions bring us much closer to authorship than asking whether a tool touched the work.
But authorship is only one side of the relationship. A work can be made brilliantly and still mean nothing to anyone. To understand why some work matters and other work disappears, we have to look at what happens on the other side.
We have to look at the audience.
What the Listener Is Really Giving
A song can exist as a file without ever becoming culturally alive.
For that to happen, somebody else has to meet it and invest something.
The listener gives time. They give attention. They give emotion, memory, money and sometimes part of their identity.
This is the part of the market that matters most. Not simply what people buy, but what they decide deserves a place in their lives.
It also explains why technical quality is not enough.
AI music can sound as good as a commercial release. If some of these songs had been released by known artists and pushed through the right channels, they might easily have become hits. But the label AI changes the listener’s willingness to invest.
Why?
Because people do not only love artists for the artists.
At a deeper level, they love what artists help them find inside themselves.
A person hears a lyric and thinks:
This person thinks like me.
This person feels what I feel.
I am not the only one.
The artist is a mirror, but also another human being standing on the other side of the feeling. The artist becomes a witness.
That is why Boney M needed Bobby Farrell. The record already had a voice, but the voice needed a body the audience could meet. Farrell made the machinery feel like a person.
An AI song can produce the mirror.
What it does not automatically provide is the witness.
Once the listener knows there may be nobody behind the voice, another question appears:
Who understood me?
This does not mean AI music can never create a real connection. A person can use AI and remain the human centre of the work. They may bring the experience, the idea, the direction, the choices, the voice and the meaning. The technology may help them express something they genuinely perceived.
The future may not divide neatly into human music and AI music.
It may divide between work that carries a believable human imprint and work that does not.
That distinction will not appear neatly or immediately. Before it settles, AI will continue breaking apart categories that once seemed stable. That period of fragmentation will feel chaotic. But it may also be how we reach a more honest kind of coherence.
From a False Whole to an Honest One
Before AI, the industry gave us a neat picture.
There was an artist at the front, and everything seemed to flow from that person. The picture felt whole because all the hidden parts had been compressed into one identity.
But it was often a false kind of wholeness.
AI breaks that picture into visible pieces. The voice can be separated from the body. The writing can be separated from the writer. The performance can be separated from the experience. Production can be separated from the producer.
For a while, that will create confusion.
There will be labels, detectors, school rules, court cases and new laws. Some protections are clearly needed. People should control the use of their voices and faces. Fraud, theft and false credit still matter.
But rules can limit harm. They cannot decide what deserves our attention.
That will still be decided through the investment of listeners, readers, viewers and users.
Over time, the scattered pieces may come back together in a more honest way. We will not need to pretend that one celebrity did the work of twenty people. We will be able to ask what the performer did, what the writer did, what the producer did, what the AI did and who gave the whole thing its direction.
That is a different kind of integration.
It does not hide the parts.
It makes their relationship clear.
Once those parts become clear, something else begins to happen. The abilities that once looked rare become common, and our attention starts moving towards the qualities the machinery cannot guarantee.
Value Moves Up the Chain
When something is scarce, we often confuse its scarcity with its value.
For a long time, polished production was scarce. Most people could not make a song that sounded ready for radio. So the polished sound itself seemed to prove that something extraordinary was behind it.
Now polished sound is becoming common.
Once everyone can reach it, polish stops proving very much.
That does not mean human taste becomes less important. It means taste becomes sharper. We stop being impressed by what has become normal and begin noticing forms of value that were harder to see before.
We start seeing in higher resolution.
In music, technical polish becomes common, so presence, taste and a believable human voice matter more. That voice may be edited or even partly synthesised, but listeners will still want to feel that it leads back to someone.
In writing, clean prose and clever language become common, so the idea matters more. Did the writer see something that was not obvious before? Is there a real structure beneath the sentences?
In programming, producing code becomes easier, so system design, judgment and responsibility matter more.
In education, explanations become endless, so knowing what a particular person needs—and helping that person grow—matters more.
The same thing happens to public intellectuals. For years, polished language, clever references and a confident public image could make thin thinking look deep. Sometimes a whole team helped maintain the image.
When everyone can produce the surface, people eventually look beneath it.
Is there an idea?
Does it hold together?
Does it explain anything?
Can it solve a problem?
AI does not ruin intellectual life.
It makes sounding intellectual ordinary.
Then we can start asking who can actually think.
That brings us back to the phone with which we began—and to the question underneath the entire essay.
What Was the Human Adding?
The phone can make a song.
It can make a very good song.
That fact forces us to ask what the human being was adding all along.
The answer was never simply the singing. Boney M already showed us that.
It was never simply having the right face. Milli Vanilli showed us how quickly that could collapse.
It was not even enough to perform a feeling convincingly. t.A.T.u. showed us that the people living the truth of a song could be the listeners rather than the performers.
What people wanted was a complete relationship between sound, meaning and a human source.
The old music industry often gave us that relationship as an illusion. Many people did the work, one person embodied it, and the audience treated the complete package as real.
AI pulls those parts apart in front of us.
At first, that feels like destruction. The voice may be artificial. The lyrics may be assisted. The instruments may never have been played. The face may not exist.
But when the old signs of value become easy to reproduce, value does not disappear.
It moves.
It moves towards perception, taste, meaning, judgment, presence and responsibility.
It moves towards the part of the work that somebody can genuinely stand behind.
AI did not suddenly make music impure. It showed us how assembled music had been for a long time. It revealed the bargain we had quietly accepted: if the sound came with a convincing human face, we were willing to experience many separate parts as one.
Now the machinery is visible.
The task is not to pretend it can be hidden again. It is to build a more honest whole around it.
AI did not ruin music.
It revealed it.
Maybe that’s not such a bad thing
