A few years ago I made the Soppoppia tracks, a tribute to computer games from the 1980s and to mushrooms (the title in Norwegian means roughly "shroom up yes"). Piano keyboard, long evenings testing synths, trimming , mixing it myself over weeks.
A few months ago I made another mushroom track, Spore & Signal, this time with AI tools. It took me about a day. I wrote prompts, chose between versions, made some edits, picked an AI mastering at the end.
Both are long instrumentals. Both put me in the same mood on a Gibralfaro walk. But I listen to the AI one more. The track I barely touched moves me more than the one I built by hand.
That bothers me. I always assumed what moved me was partly the human struggle behind a thing. Here there was almost none, and it moved me more.
Maybe the unease rests on a line that was never as clean as it sounds. The line between "made by a human" and "made by a machine."
Electronic music has been tool-mediated for decades. Autotune. Sampling. Multi-track recording. Tape loops. Producers who did half the actual creative work and barely got their name on it.
The same is true of books you have read. My Struggle is credited to Karl Ove Knausgård; his editor Geir Gulliksen worked so closely on the six volumes that the title itself — the deliberately provocative Min Kamp, the same words as Mein Kampf — was Gulliksen's idea, against Knausgård's initial resistance. Pedro Almodóvar's films are credited to him; José Luis Alcaine often shot them, Alberto Iglesias scored most of them, his brother Agustín Almodóvar produced almost all of them.
The names that go on the cover are a convention, not a complete fact.
AI is not a clean break from that history. It is the steepest step on a gradient that was already there.
And the steepness matters. The hour you spent fighting with a synth until it sounded right, the afternoon you spent rewriting a paragraph until it stopped sounding wrong, the year you spent learning what a chord progression actually felt like. That friction was not just inefficiency. It was where taste was slowly being built, as a byproduct of doing the work. The thing still exists when the friction disappears. But the maker is not built the same way.
We say we care about the finished song, the finished novel, the finished painting. But part of what moves us is the implied human struggle inside the thing. The sense that another nervous system spent time wrestling something into existence.
If a song generated in a few hours moves me more than one I built slowly over weeks, what exactly am I responding to? The sound itself? The mood? Some pattern my brain likes regardless of where it came from? Or had I quietly assumed that effort and meaning were more connected than they really are?
So what is left for the maker, when execution becomes cheap?
Taste, perhaps. Knowing what is good and what is not. Knowing which of fifty versions to keep, often without being able to say why. Knowing when a sentence sounds plausible but is not quite right. Knowing when a chord change is carrying emotion. Knowing when a paragraph is finished and when it still needs work. Tools amplify decisions. The decisions are still yours.
I write these essays with an AI in the loop now, drafting, cutting, arguing back. I often cannot tell which sentences started as mine, which came from the machine, and which became mine the moment I decided to keep them. The 'good' and the 'mine' have folded into each other, and that confusion is the whole problem in miniature. It is also not just mine. Anyone using these tools is going to bump into the same question.
If friction is where taste is built, and AI removes friction, then the people developing taste now may be running on capital accumulated by older makers who had to struggle. What replaces that, once struggle is optional, I genuinely don't know.
A large share of modern professional life has rewarded reliable execution. Writing the deck. Building the model. Learning the software. Producing on schedule. Now many of those tasks are becoming cheap. The remaining difference between people moves toward what is harder to copy. Judgement. Selection. Taste.
Taste is hard to develop because the feedback loop is slow. You make something, you put it out, no one tells you if it is good or just competent. You guess. AI removed part of the cover that execution used to provide. Competence used to read more easily as depth because competence itself was scarce. Now competence is everywhere.
There is a side-effect worth naming. When execution was the bottleneck, our capitalistic society valued brains that could focus, plan, and produce on schedule. People who thought sideways (restless attention, narrow obsessions, the kind of mind that fixates, that sees connections others miss, that refuses the obvious answer) were the exception, often viewed as a problem. When taste is the bottleneck, the balance inverts. The brain that was good at producing-what-was-asked has its job done by machines. The brain that fixates, that thinks sideways, becomes the interesting one. None of this cures the difficulty of living with ADHD or autism. But the unusualness itself has a value it did not have before. The outlier brain is not the bug it used to be.
Maybe in ten years AI work will just be "work," the way electronic music is now just "music." The asymmetry I feel now may not survive for long. What probably will survive is taste. And perhaps something even harder to measure than taste: sincerity. The faint but detectable difference between something made by someone trying to say something real, and something generated by someone trying to occupy space.
Art was once evidence of someone struggling toward something. The mind's role has shrunk toward taste and selection. A great deal of what gets called "good art" in the next ten years will be people with strong opinions, confidently shipping polished work without sincerity. I am not sure I am ready for that. I still look for the human struggle in the things I love.
Of course, "good taste" was never stable. What the crowd likes, what critics canonise, what you quietly prefer, what you liked as a child before you were taught what to admire. AI just makes the question harder to avoid, now that the technical barriers to making things have mostly fallen.
Whose song is it? Mine, in the sense that I chose it. Mine, in the sense that I kept it. Not mine, in the way I used to mean when I said 'mine.' We are between two definitions of ownership, and everyday language has not caught up.
If you learned an AI wrote that book or song you love, would you still feel the same way?