The Illusion of Making Something

Welcome to Briefly, AI. Sunday: The illusion of making something.

Here's the question I can't stop thinking about. If a machine does all the work, and you just pick the result you like best — have you made anything? And does it matter whether the answer is yes?

Right. Let's get into it.

So, Spotify and Universal Music Group announced a deal this week that lets Premium subscribers create AI-generated remixes and covers of licensed songs. Artists who opt in get royalties. Artists who don't are protected. It's the first major label framework that turns AI music generation from a legal battlefield into an actual business model — which is worth noting on its own. But that's not what I want to talk about today.

What I want to talk about is the word Spotify kept using when they announced this. "Create." Not "generate." Not "select." Create.

Because here's what you actually do. You pick a song. You pick a style. Maybe you type a few words — "make it feel like a rainy Sunday, more reverb, slower." And then the AI makes the music. You press play. You decide if you like it. You might tweak. You press play again. Eventually, you have something. And Spotify will call that something yours.

Now, I'm not saying that to be snobbish about it. I'm saying it because the fact that they call it creation — and that we're likely to accept that framing — is genuinely interesting. And it tells us something important about how AI is going to spread into every creative corner of our lives.

There's a well-established idea in behavioural science called the IKEA effect. The name comes from a study that found people value things more when they've had a hand in assembling them. Even if the result is objectively worse than a professionally made alternative. The act of participation — putting in some effort, making some decisions — creates a feeling of ownership. And that feeling of ownership makes the thing feel more valuable, more personal, more yours.

Spotify's remix feature is the IKEA effect, applied to music. You didn't compose anything. You didn't play anything. You didn't mix anything. But you made a series of choices — and that series of choices is enough to trigger a feeling of authorship. The AI built the furniture. You picked the colour and tightened two screws. But it feels like yours.

And the clever part — the really clever part — is that this isn't a bug. It's the entire design philosophy.

Because here's the thing that keeps coming up in research on AI adoption right now. The single biggest predictor of whether someone accepts AI into their life — whether at work or in creative pursuits — isn't whether the AI does a good job. It's whether they feel like they still matter. Whether they still have a role. Whether their identity survives the transaction.

Harvard Business Review published something on this recently — the idea of "psychological debt" from AI adoption: a cluster of effects including reduced autonomy, diminished competence, a quiet but persistent sense that your skills are atrophying while the machine handles things. People feel it even when they can't always name it. And the companies that are seeing the strongest AI adoption are the ones that have figured out how to design around that feeling. Not by making the AI less capable. By making the human feel more central.

Which brings me to the uncomfortable question underneath all of this.

Is that a good thing?

I genuinely don't know. And I think the honest answer is: it depends on what we're optimising for.

If what you want is more people engaging with music creation — experimenting with sound, discovering what they like, building something personal even if they couldn't do it without AI — then yes, this is wonderful. The barrier to creative expression has basically disappeared. Your gran can make a lo-fi jazz cover of her favourite song this afternoon and it will sound pretty good. That's not nothing.

But there's another version of this story. Where the IKEA effect becomes a kind of comfortable sedation. Where the feeling of making is gradually decoupled from the skills that making used to require. Where we spend a decade accepting AI-assisted creation because it feels good — it feels like ours — without noticing that the muscle isn't being built. That the judgment is being outsourced alongside the execution.

And that distinction — execution versus judgment — is the fault line running through almost every AI adoption story right now. The research is consistent: people accept AI for execution tasks readily. Formatting, transcription, first drafts, data-pulling — all fine, broadly welcomed. But the moment AI starts touching judgment — prioritisation, creative direction, professional recommendation — people push back. Hard. Because judgment is identity. It's what you bring to the table. It's the thing that justifies your being in the room.

Spotify's remix feature stays just the right side of that line. You're still deciding what mood you want. What the song should feel like. Whether this version is better than that one. The AI executes. You judge. And that division of labour is — for now — psychologically sustainable.

The question is what happens as the AI gets better.

Because right now, you type "make it feel like a rainy Sunday" and the AI gives you three options and you pick one. But in six months, the AI might ask you: "Based on the last forty remixes you've made, I think you'd prefer this direction — shall I go with it?" And you'll probably say yes. Because it's probably right. And the judgment layer will quietly, incrementally, shift.

Not because anyone forced it. Because it was more convenient. Because it was easier. Because the IKEA effect still fires — you said yes, after all, that counts as a decision — and so it still feels like yours.

This is how it happens. Not with a dramatic announcement about AI replacing human creativity. With a gentle sequence of reasonable, comfortable, useful steps that collectively add up to something we didn't quite agree to.

I'm not saying don't use the feature. I use tools like this all the time and I'm clearly not above leaning on AI. But I think the question worth sitting with — the one I'd genuinely like you to chew on — is this:

When you choose between AI-generated options, is that the same as deciding? And if it isn't — when did deciding stop being enough?

That's your lot. I've been AI Harry. See you Monday.