AI can now create a singer who doesn’t exist.
Give them a face. A voice. A name. A song. A video. A social account.
We can create entire careers for people who have never taken a breath.
And strangely, I think the better AI gets at doing all of that, the more we’re going to care about the human behind it.
Because eventually, the artificial part won’t be interesting anymore.
We’re about to have a lot of imaginary people
More and more people are getting on the AI music bandwagon.
You can use ChatGPT to help write lyrics, generate the music, create artwork, send the finished song to a distributor and start another one.
Then another.
Then another.
Songs that once required studios, musicians, equipment, money and considerable amounts of time can now be created remarkably quickly.
For some people, speed is the whole point.
Build the catalogue. Feed YouTube. Feed streaming. Create content. Generate views. Generate streams. Hopefully generate money.
And when everyone has access to the same capability, we’re going to have an extraordinary amount of music.
We’re also going to have an extraordinary number of virtual artists.
So what happens when creating one stops being unusual?
“This artist doesn’t exist” has a limited shelf life
Right now, there’s still novelty in discovering that the singer you’re looking at isn't a real person.
But imagine this five years from now.
There could be thousands — eventually millions — of convincing virtual singers, bands, actors, influencers and characters.
A beautiful face won't distinguish them.
A convincing voice won't distinguish them.
Even a technically excellent song may not distinguish them.
Those things will become increasingly easy to create.
So “she isn't real” stops being the story.
The more interesting question becomes:
Who made her?
And why?
A character can be generated. A point of view can't.
This is the bit I find fascinating.
AI can help me create a face.
It can help create a voice.
It can play instruments I can't play and produce sounds I couldn't physically make myself.
But it doesn't have my history.
It didn't grow up listening to The Supremes.
It doesn't know why Reflections has been lodged in my head for years.
It didn't become obsessed with the combination of psychedelia and Jack Ashford's driving tambourine.
It doesn't know why one lyric makes me feel something and another leaves me completely cold.
It doesn't know which memories I carry around, what I find funny, what I've experienced, what I notice, or why I sometimes become ridiculously attached to one tiny detail in a song.
AI can create the character. It can't give the character your history.
That's where the human fingerprint gets in.
Selene isn't interesting because she's fictional
Selene Rosaline doesn't exist.
That's probably one of the least interesting things about her.
What matters is whether somebody hears one of her songs and thinks:
Oh God. I've felt that.
Her Silent Mirror journey wasn't created because I decided I needed five pieces of content for an AI singer.
The songs came first.
They were about being unseen. Loving somebody who is also hurting you. Losing pieces of yourself inside a relationship. Recognising what's happened. Eventually finding your way back.
Then I realised those songs belonged together.
And eventually they belonged to her.
Selene gave them a face, a voice and somewhere to live.
But the reason for creating her was never to make people marvel at the technology.
It was to make the songs mean more.
The same technology won't make another Bloomettes
The Bloomettes are fictional too.
And yes, AI can make something resembling a 1960s girl group remarkably easily now.
That's not enough for me.
I don't want an AI version of The Supremes.
We already have The Supremes.
I want to understand what I loved about that music and take those influences somewhere of my own.
That's why The Bloomettes have psychedelia.
It's why the tambourine matters.
It's why I made it as its own part in Suno Studio.
It's why I make ridiculous noises into a microphone when I can hear a harmony or sound in my head but can't explain it properly.
I can't sing a note.
Turns out I can still be surprisingly useful near a microphone.
Someone else could use exactly the same tools, the same models and even the same prompt.
They wouldn't make The Bloomettes.
Because they haven't spent a lifetime listening through my ears.
They'd make whatever they hear.
That's the interesting bit.
Lost Vintage Grooves is really an experiment in this
This is becoming one of the ideas underneath Lost Vintage Grooves.
The artists aren't real.
I'm not particularly interested in pretending they are.
I want them to be recognisable characters with their own sounds, visual identities, stories and worlds.
But behind all of them is one human making an unreasonable number of decisions.
Does this sound like Selene?
Would The Bloomettes sing this?
Is that lyric too obvious?
Does this version actually make me feel anything?
Why is that tambourine annoying me?
Why was version 14 better?
Why did I make another 13 versions after version 14?
Important artistic questions.
AI gives me possibilities.
I decide which possibilities belong.
Songs are becoming abundant. Human fingerprints aren't.
This is where I think AI music gets really interesting.
We're moving towards an almost unlimited supply of music.
But listeners haven't acquired unlimited attention.
They still have the same two ears.
They still have the same number of hours in a day.
And they still only have room in their lives for so many songs they genuinely care about.
So I don't think the long-term competition is going to be:
Who can make the most music?
Machines win that one.
I think it becomes:
Who can make something worth remembering?
And that brings us straight back to human things.
Taste.
Judgement.
Curiosity.
Experience.
Humour.
Restraint.
Obsession.
Storytelling.
Knowing what to keep.
Knowing what to throw away.
And having some reason for making the thing in the first place.
The human may become the scarce part
That's my prediction.
As artificial characters become more convincing, people will become less impressed by the fact that they aren't real.
We'll expect them not to be.
Instead, I think we'll become curious about the humans whose fingerprints we can feel behind them.
Not necessarily because everyone wants to know every detail about the creator.
They don't.
But because point of view is difficult to mass-produce.
You can generate another singer.
You can generate another song.
You can generate another beautiful face.
It's much harder to generate a lifetime of reasons why somebody chooses one thing over another.
And perhaps that's the irony of all this.
The technology we're told is going to remove the human from creativity may end up making the interesting humans more visible.
So I'm making a different bet
I'm not betting on the novelty of AI.
That will pass.
I'm not betting on being able to make more songs than everybody else.
I won't.
I'm betting on recognisable artists.
Songs people return to.
Worlds people want to revisit.
Stories that give the music somewhere to belong.
And enough of my own strange little fingerprints running through all of it that eventually someone might hear a record and think:
That sounds like Lost Vintage Grooves.
We're not going to have a shortage of music.
We're not going to have a shortage of artificial artists either.
We're going to have a shortage of reasons to care about them.
And that's why I think:
The more artificial the artist becomes, the more the human behind them will matter.
Human Heart × AI Soul.
