The Case Against Taste

As AI gets better, personal taste is losing value #

As AI tools improve, the need for human creative judgement changes. To see why, let’s start with the job of a creative director. What is a creative director? They lead teams of creatives, shape creative vision, and ensure a standard of quality.

Up to now, AI hasn’t even acted as a junior creative. Generative AI tools have operated under our direction, producing mostly inconsistent work with flashes of brilliance. Nothing surprising. Nothing original. We have been their creative directors, assigning work to the models, using our judgement to assess outputs, and pushing them with prompts until something worked.

Today, our creative judgement adds value to AI. We decide what is art and what is slop. Models can write scripts, generate images and videos, and compose music. But, only we can look at the outputs and judge if they are good.

What happens when AI gets a promotion? #

I’ve talked in the past about the Cherry Pick Ratio: the number of outputs an AI model has to generate before you get a useful output. That ratio is rapidly approaching 1:1. And when AI can generate a useful creative output every single time, our ability to judge the quality of those outputs suddenly becomes less valuable. More than that, when we layer in agentic creative workflows, AI not only becomes a trusted, senior creative; it becomes the creative director.

The role of humans in creative collaboration will change. And we need to evolve our role when working with generative AI systems.

We developed creative judgement because making things was traditionally expensive. Every frame of every film, every sketch, every prototype cost real resources. Judgement was an economic filter as much as an artistic one. We had to be selective because waste was unaffordable. We could only spend so much time, energy, and money on a project.

AI removes this limit. It can generate creative faster and cheaper than ever before. Entire creative industries no longer rely on our judgement. AI is already generating stock images, elevator music, and X posts with little oversight. Further upmarket, judgement is still at the centre of creative work seen as more artistic or high-value, like feature films or fine art. These industries may be an order of magnitude more difficult. But, as the generative Cherry Pick Ratio approaches 1:1, our relationship to creativity will need to evolve.

Why we still matter #

None of this is an argument against human creativity. Rather, it’s an argument that we’ve been defending the wrong thing. What is the unique value that we add to the co-creation equation?

The blunt truth is that creative judgement is no longer only a uniquely human skill. AI models can now perceive, critique, and comment on multimedia with something close to a perspective. Taste has become infrastructure. It is embedded in tools and outputs. It is no longer a function creatives perform but something they build on.

But, the model bears no consequence. It cannot stand behind its work and ask: “Do I own this? Am I proud to share it?” It sacrifices nothing to create.

We don’t matter because we have good judgement or can conjure outputs with context, prompts, and preambles. We matter because we are finite. We matter because we end. A machine can generate a masterpiece, but it risks nothing to do so. The case against taste is simple: we don’t need humans to tell us what is good anymore. We need humans to tell us what is worth the risk.

Our edge isn’t our minds. It’s our humanity.