Taste Cannot Be Prompted

There is a version of this argument that says AI will replace the artist. It won't. The artist is not the person who makes the sound. The artist is the person who decides which sound is worth keeping.

Every tool accelerates production. The loom didn't replace the weaver's judgment about what to weave. The synthesizer didn't replace the musician's ear. AI is the latest acceleration. It is not the latest replacement.

What changes is the ratio. One person with sufficient taste can now produce at a volume that previously required a team. That is not democratization. That is concentration. The output expands. The selection problem sharpens.

Prompting Records exists on that edge. We use AI to collapse the distance between the idea and the rendered sound. We do not use it to skip the decision about whether the sound is good.

That decision is not a feature. It cannot be added to a model. It is the product of years of listening — to what works, to what fails, and to the silence between those two things.

That question — what separates AI-native from AI-only — is one worth holding separately.

Taste cannot be prompted.