This article was written by AI

Ok, so it wasn’t. But it was read, a myriad of grammatical errors identified and changed, with some hurtful criticism of meandering ideas and sentences (hopefully now resolved!). But do I get a pass for admitting it?

AI warnings are becoming a real thing — and not in a hypothetical, future-facing way. Right now.

Recently, the BBC reported on how AI-generated music is quietly percolating into Spotify’s ecosystem, where “Made by Human” and “Made by Machine” tracks increasingly sit side by side. Unsurprisingly, some listeners aren’t happy about it.

We shouldn’t be surprised.

 At a very fundamental level, people don’t just value outputs — they value effort. More precisely, they use perceived labour as a proxy for authenticity, care, and connection to a creator. This shows up everywhere. Why do you think people pay more for artisan foods. We seek out handmade materials. We linger over handwritten labels in the store, in a way we simply don’t with printed ones. Labour, especially human labour, signals meaning in a way that artificial labour can’t.

There’s a slightly counterintuitive twist here, and it’s one I have explored in my own research. In a series of experiments looking at digital products (think NFT-style assets), we found that when a physical equivalent of the same product was introduced, evaluations of the digital version dropped sharply. Nothing about the digital product changed — except its comparative proximity to something more “real,” more effortful, more human.

This is the backdrop to the current unease around AI-generated music. In response, independent developers have already created filters that identify “made by machine” tracks — and they’re gaining traction. Which raises an obvious question: should Spotify introduce its own version? A simple label. A transparency move. A warning of sorts…

It sounds sensible (you would, right). But it’s not that simple.

Spotify doesn’t just distribute music — it sells access to it, while carefully avoiding the perception that it curates taste. Introduce an AI label, and perhaps they cross that line. They move from being a neutral platform to an implicit arbiter of what counts as “authentic,” or worse, what counts as “good”. I have no doubt this idea is unpalatable to executives. I mean, their entire value proposition is built on fitting preferences, not shaping them overtly.

To understand the problem properly, we need to step back and think at the level of the product or service itself. AI warnings don’t have a uniform effect. But they do reallocate value, depending on what consumers are actually looking to purchase.

Consider this.

 For functional, low-empathy, efficiency-driven categories — insurance comparisons, travel aggregation, finding the cheapest coffee cup — a “made by machine” label is not a warning. It’s a signal of optimisation. It says: this has been generated at scale, processed quickly, and stripped of unnecessary friction for your benefit. In these contexts, AI enhances value. Warnings become an asset to be positioned.

But move to expressive, high-empathy categories — music, art, literature, many types of food — and the same label does something very different. It weakens the perceived link between creator and creation. And in doing so, it strips away a key source of value people seek (and ironically one they took for granted in many categories in the past).

That is Spotify’s dilemma.

And here’s the twist.

 The long-term health of platforms like Spotify may depend on keeping music just non-commodity enough to matter. Not fully optimised. Not fully flattened. Not fully machine-made — or at least, not perceived that way.

 AI warnings, then, are not simply about transparency. They are about where value comes from.

 Get that wrong, and you don’t just label the product differently.

 You change what it is! Dangerous.