There's an AI-generated country song at the top of the charts. I keep playing it, partly because everyone is talking about it, partly because it's genuinely good. Not unforgettable, not haunting, just competently polished in that modern radio-ready way. Warm guitar, confident rhythm and a chorus with enough swagger to feel familiar. And then five minutes later try humming it and it vanishes completely.
That forgettability should bother me more than it does. Because last month, I made a similar choice. I needed a header image. I could have commissioned an illustrator for £150, maybe £200. Instead: Google’s image model. Ninety seconds. Four options and < £0.05. The uncomfortable part was how quickly I accepted it. I looked at the image, thought this is good enough, and only later felt the sting of what “good enough” had cost someone else. I don’t know who I didn’t hire. But I know I’m part of this shift. And I know I’ll do it again. Once barriers collapse, you can’t pretend they’re still standing.
The backlash around “Walk My Walk” - the AI track that hit #1 with just 2,500 downloads - reveals more about us than the model that made it. Fans weren’t angry because it sounded bad. They were angry because it skipped the suffering the genre romanticises: no dive bars, no heartbreak, no hungry years. But if most listeners couldn’t tell until they were told. What exactly are we defending - the art, or the narrative? Perhaps the answer lies in a quality we rarely discuss: a song's ability to be truly unforgettable. In this new landscape, that very quality is under threat.
Indeed, AI seems destined to dominate 'good enough' music – tunes so frictionless you don't turn them off while driving. This leaves humans competing for the shrinking space of the 'unforgettable.' To better understand this distinction and how AI itself perceives musical quality, I decided to put it to the test. I asked AI to critique the song, and its analysis was striking. It precisely identified complex techniques like 'False Chord Grit' and 'Gospel Stacking' – skills it estimated would take a human 8–12 years to master.
It is hard to feel the weight of that disparity until you see it on paper. So I mapped the two realities onto receipts to see exactly what is changing in the creative transaction.
On the left, the Human Cost. On the right, the AI Reality.
Both receipts produce a song, but only one requires a decade of invisible human investment
This is the shift. The Songwriting, the actual intent and direction, remains the constant. The machine didn't replace the idea. But look at the first section: Prerequisites. In the old world, you paid a "Time Tax" of 12 years just to be allowed to enter the studio. In the new world, the Time Tax is waived. Access is granted instantly.
Many artists are trying to adapt - taking AI courses, rebuilding portfolios, but refinement pays a fraction of what creation once did and the pool keeps shrinking. More than a few have taken part-time jobs just to stay afloat. Not because they’re unwilling to evolve, but because the shift is happening faster than any individual can realistically keep up with. What’s emerging isn’t “AI versus humans.” It’s a creative world splitting into three tiers and the speed at which you move between them depends on resources most people don’t have.
Tier One is the Hybrid Winners: the illustrators thriving because they’re bending with AI. One prototypes fifty concepts before breakfast, then hand-finishes the best three. Another layers texture and imperfection onto AI-generated album art until it feels made rather than rendered. Their competition isn’t the machine - it’s other humans who use machines better.
Tier Two is the Struggling Middle. Across illustrator forums and industry surveys, the story barely changes: full commissions are disappearing, replaced by low-paid “AI refinement” work. A UK survey found 26% of illustrators have already lost jobs to AI, and over a third say fees have fallen as £600 - £1,000 covers turn into £150 cleanup gigs. Reuters and The Guardian report the same shift - long-standing clients swapping creation for correction because it’s cheaper.
Tier Three is Automation Alone: the AI song, my image header, the flood of competent content-eating work no one wanted to admit was fragile; it's the “nice to have” illustrations, background tracks, placeholder designs that become permanent because no one wants to spend more.
Adaptation isn’t willpower, It's a runway. Time, space, stability. The hybrid winners had established clients and the breathing room to experiment. The ones drowning didn’t lack talent - they lacked oxygen. And once you’re behind, the gap accelerates. This isn’t automation replacing humans. It’s automation stratifying them.
However, this stratification isn't the whole story; for many, AI is proving to be a powerful equaliser, opening doors that were once firmly shut.
And yet, somewhere right now, a 17-year-old in rural Rajasthan uploads her first song. No teacher, no studio, no industry contact. She creates a track that sounds professional enough to sit on a playlist next to a major label artist. Twenty years ago, her talent would have died unheard because she couldn't pay the entry fee. An illustrator with arthritis can draw again. A choir teacher builds custom backing tracks for a school with no budget. These aren’t edge cases - they’re creative lives expanded in ways unimaginable a decade ago. Two contradictory realities. Both are true and both are urgent.
We are living through the most significant creative shift since the printing press and treating it like a software update. Nothing about this trajectory is predetermined. We still have time to build guardrails.
What do we protect?
Transparent attribution: If an AI song hits #1, we should know it. Not because AI art is lesser, but because we need visibility into what’s happening. We can't study an experiment of this scale if we can’t see its inputs.
Fair compensation: For those whose work trained the models. If millions of artists’ and musicians’ work shaped the intelligence of these tools, they’re not “inspiration.” They’re suppliers.
Economic oxygen for adaptation: The hybrid winners weren’t magically talented. If we want people to transition instead of being discarded, we need to give them time. Short-term income support. Scholarships. Retraining grants tied to real outcomes.
These aren’t anti-AI positions. They’re hopeful ones - because they assume a future where humans and machines create together, not at each other’s expense.
The AI song is still playing. I still can’t hum it. But I remember the songs that required struggle. I want my niece to grow up in a world where the people whose work trained her tools are valued, not forgotten, and where AI expands creativity instead of hollowing it out. We won’t understand the full consequences for some years. But there is still time to choose a future where humans and AI rise together.
The machine didn't steal 12 years of learning. It gave us 12 years of runway. We can use that time to flood the world with more 'good enough' content. Or we can use it to bypass the technical grind and go straight to the human core - to tell stories that were previously too expensive, too difficult, or too complex to exist.
We have been given a miracle of efficiency.
The real question isn't whether AI will replace the artist. It is: What will the artist build with all this new time?
