What unsettles me is not Sarah Thomson. Or Michael Bennett.
It is the mirror.
I discovered two singers who do not exist, yet they moved me.
Sarah’s face feels trustworthy. Her voice feels lived-in. Her songs feel personal. Her performances seem authentic.
Then someone tells me that Sarah Thomson is not a woman who wakes up in the morning, burns toast, regrets old decisions, misses dead relatives, or worries about paying taxes. She is an artificial construction. A manufactured ghost.
And suddenly the floor tilts.
Because if a person who never lived can make us feel understood, what exactly have we been responding to all these years?
That question is bigger than music.
It is a question about art, humanity, memory, identity, economics, psychology, and perhaps even consciousness itself.
Sarah Thomson and Michael Bennett appear to be part of a growing wave of AI-generated music and film projects. Some platforms and tracking services have explicitly flagged their releases as AI-generated. Sarah’s character is presented as a seventy-year-old mother telling stories about faith, loss, regret, and family.
The strange thing is that many listeners report exactly the same I felt.
They trust her.
They believe her.
They feel something.
And that is where things become interesting.
The First Lie
People often say:
“AI created this song.”
No.
Not really.
The truth is simultaneously less impressive and more terrifying.
AI did not wake up one morning and decided to write songs about motherhood.
Humans built the system.
Humans trained the system.
Humans selected the style.
Humans chose the themes.
Humans chose which output survived.
Humans uploaded the final version.
Sarah Thomson is not AI creativity.
She is human creativity filtered through AI machinery.
The paintbrush did not paint itself.
The piano never composed Beethoven.
The camera never photographed Paris.
The tool changed.
The artist did not disappear.
At least not yet.
This distinction matters because people often compare AI and humans as if they are competitors standing in opposite corners of a boxing ring.
The reality is much stranger.
AI is built from us.
Every song it generates emerges from patterns extracted from human creations.
It is humanity remixing itself at impossible speed.
A gigantic cultural photocopier with increasingly sophisticated judgment.
The question is therefore not:
“Is AI more creative than humans?”
The question is:
“Can humanity become more creative by building machines that rearrange humanity?”
That answer is increasingly looking like yes.
The Second Lie
People also say:
“AI doesn’t understand emotions.”
That depends on what you mean by understand.
Does Sarah Thomson feel grief?
No.
Does she miss a son who moved away?
No.
Has she stood beside a hospital bed?
No.
Has she buried a husband?
No.
Yet her songs can still produce genuine grief in listeners.
This reveals something uncomfortable.
Art has never required the artist’s suffering.
It only required the one of the audiences.
A novel can make you cry even though the characters never existed.
A film can terrify you even though the monster was made of latex.
A painting can break your heart despite being only a pigment on canvas.
The emotional event happens inside the audience.
Not inside the artwork.
Not inside the artist.
Inside you. Inside me. Inside us.
That realisation strips away one of our favourite romantic myths.
We like believing that every great song is a diary entry.
Sometimes it is.
Sometimes it is a team of songwriters in a conference room manufacturing heartbreak before lunch.
Human beings have always industrialised emotion.
AI simply makes the factory more visible.
Why Sarah Thomson Feels So Perfect
I noticed something important.
She seems almost impossibly likeable.
Natural.
Trustworthy.
Comforting.
Authentic.
That is not accidental.
It may be the result of optimisation.
Human performers are messy.
They have bad interviews.
Bad opinions.
Bad divorces.
Drug problems.
Political opinions.
Tax scandals.
Arrogance.
Egos.
Entire careers dedicated to self-destruction.
Humans are magnificent and exhausting.
Sarah Thomson has none of these issues.
She exists in permanent alignment with audience expectations.
No awkward press conference.
No drunken rant.
No terrible haircut phase.
No ageing badly under paparazzi lenses.
No disappointing autobiography.
She is pure signal with almost no noise.
The entertainment industry has spent a century trying to manufacture stars.
AI may simply be the next stage of that process.
The perfectly calibrated performer.
Not necessarily because AI is creative.
Because AI is editable.
That difference matters.
Does Generative AI Know Us Better Than We Know Ourselves?
Possibly.
At least statistically.
Which sounds less romantic and more horrifying.
A human songwriter might know fifty people.
A hundred.
A thousand.
An AI system can be trained on millions of examples of what people listen to, skip, replay, save, share, cry to, dance to, and remember.
It sees patterns no individual human can see.
Not because it is wise.
Because it is enormous.
Humans understand stories.
Machines understand distributions.
Both are powerful.
A songwriter may know why heartbreak hurts.
An AI may know precisely which chord progression maximises listener retention after twenty-three seconds.
One understands meaning.
The other understands behaviour.
The future belongs to whoever combines both.
The Netflix Problem
Sarah Thomson is not really a music problem.
She is a symptom of a much larger phenomenon.
Look at modern entertainment.
Streaming platforms already optimise content around audience preferences.
Algorithms recommend books.
Recommend films.
Recommend music.
Recommend friends.
Recommend lovers.
Recommend political opinions.
Recommend outrage.
Recommend happiness.
Recommend identities.
Human culture is increasingly mediated through prediction.
Sarah Thomson may simply be the logical next step.
Instead of recommending music…
The system generates it.
Instead of finding a singer…
The system creates one.
Instead of discovering talent…
The system manufactures talent.
This sounds futuristic.
Actually, it sounds like marketing departments are finally getting everything they ever wanted.
A sentence so depressing it should probably come with a government warning.
Are Human Performers Finished?
No.
But their role is changing.
People often assume technology replaces occupations.
Usually, it changes them.
Photography did not kill painting.
Film did not kill the theatre.
Television did not kill radio.
E-books did not kill printed books.
The internet did not kill writers.
Unfortunately for all of us, writers survived and continued producing opinions…
Human performers still possess something AI lacks.
A real life.
When people attend a concert by Bruce Springsteen or Taylor Swift, they are not merely hearing songs.
They are witnessing a biography.
A history.
A person.
The performance contains risk.
Anything can happen.
That unpredictability is valuable.
Humans do not merely consume art.
They consume artists.
They want stories.
They want personalities.
They want scandals.
They want redemption arcs.
They want mythology.
Sarah Thomson may produce songs.
She cannot truly live one.
Not yet.
The Strange Future of Fame
Imagine two singers.
The first is real.
The second is AI.
The AI singer releases a flawless song every week.
Never ages.
Never complains.
Never cancels tours.
Never gets tired.
Never dies.
The human singer releases music more slowly.
Makes mistakes.
Changes style.
Experiments.
Fails.
Recovers.
Learns.
Who wins?
The answer may depend on what audiences value.
If people want perfect products, AI wins.
If people want human stories, humans survive.
The fascinating possibility is that both markets grow.
Human artists may become more valuable precisely because they are human.
Imperfection could become a luxury good.
Imagine that.
Humanity, eventually marketed as a premium feature.
What Is Creativity Anyway?
This is the question beneath all the others.
Ask ten people to define creativity, and you will receive eleven answers.
One definition is:
Creativity is producing something new.
If that is the definition, AI is already creative.
Another definition:
Creativity requires intention.
Then AI is not creative.
Another:
Creativity requires consciousness.
Then AI is not creative.
Another:
Creativity requires emotional experience.
Then AI is not creative.
The debate continues because nobody has agreed on what creativity actually is.
Humans love arguing about definitions.
Entire academic careers have been built upon it. The species occasionally resembles a group project that escaped supervision.
My view is this:
AI demonstrates creative behaviour.
Humans possess creative experience.
Those are not the same thing.
The output may look similar.
The underlying reality remains radically different.
The Writer’s Terror
Since I pretend to be a writer, there is another layer here.
One that hurts more.
If Sarah Thomson can move people without existing…
What happens to writers?
Poets?
Novelists?
Screenwriters?
The fear is obvious.
A machine can now generate stories.
Characters.
Poems.
Novels.
Songs.
Entire worlds.
But I suspect the deeper fear is not replacement.
It is a comparison.
What if readers cannot tell the difference?
Research already shows people struggle to identify AI-generated voices with reliability.
That uncertainty gnaws at creators.
Not because they fear machines.
Because they fear audiences.
The nightmare is not that AI becomes good.
The nightmare is discovering that good was easier than we thought.
Yet Something Remains
Despite all this, I think there is a reason people still travel across countries to see a performer standing under stage lights.
There is a reason people treasure handwritten letters.
There is a reason original paintings sell for fortunes.
There is a reason readers still care about authors.
Human beings are not merely pattern-recognition machines.
We are relationship machines.
Meaning machines.
Narrative machines.
When you learn that a song came from a real widow, a real veteran, a real addict, a real immigrant, a real mother, the song changes.
Not acoustically.
Emotionally.
Context becomes part of the artwork.
Sarah Thomson can simulate context.
A human artist possesses it.
That distinction still matters.
Whether it always will is another question.
The Most Disturbing Possibility
The most disturbing possibility is not that AI art becomes better than human art.
It is that AI reveals how much of art was already engineering.
How much of beauty was pattern.
How much of emotion was structure.
How much of storytelling was architecture.
Perhaps the machine is not exposing its own capabilities.
Perhaps it is exposing ours.
The magician’s greatest fear is not another magician.
It is somebody explaining the trick.
And AI may be explaining a few tricks we preferred to keep mysterious.
The Actual Irony
Sarah Thomson does not exist.
Yet she provoked this entire philosophical avalanche.
She made me question creativity.
Human uniqueness.
The future of music.
The future of art.
The future of performers.
The future of yourself.
That is a remarkable achievement for someone who never drew a breath.
The actual irony is that Sarah Thomson may be proof not of artificial intelligence’s greatness, but the greatness of humanity.
Every melody, lyric, expression, gesture, and emotional cue within her ultimately traces back to countless real human beings whose lives became data, whose art became patterns, whose emotions became training material.
She is a ghost stitched together from fragments of humanity.
A mosaic made from millions of tiny human fingerprints.
Then, when you listen to her and feel something, you are not necessarily hearing a machine.
You may be hearing civilisation talking back to itself.
A choir of the living and the dead, compressed into algorithms.
Beautiful.
Impressive.
Slightly creepy.
Very human, despite every effort to make it otherwise.
And perhaps that is the strangest conclusion of all:
The more convincing these artificial artists become, the less they tell us about machines, and the more they reveal about us. Every flawless note, every trustworthy smile, every perfectly judged lyric is another clue that humanity has spent centuries unconsciously teaching itself how to manufacture emotion.
Now the machines have learned the lesson.
The real question is whether we have.
Before you go, explore these related ideas:
The Genre Librarian With a Bad Attitude
AI Learning and Ethics: What AI Actually Does When It “Learns”
Rethinking Authorship in the Age of Technology
A quiet chamber behind the visible world.
We value your trust!
Your address is safe in the vault.
We’ll only use it to send you letters worth opening; no spells, no spam, no secret salesmen.
