Your Taste Is Your Moat
DJs figured this out decades ago.
When sampling technology exploded in the 80s and 90s, the music industry had the same fight we’re having now about AI. Culture built on the past. The past trying to control the future. Copyright battles. Accusations of theft. Predictions that “real musicians” were finished.
The DJs who thrived weren’t the ones who could sample the most. They were the ones with taste. The ones who could hear a 4-bar loop buried in a 1972 funk record and know—somehow know—it would work. They could find the needle in the haystack and recognize it as gold.
They weren’t called DJs at first. They were called selectors.
That distinction matters more than ever.
Generation Is Cheap. Selection Is Not.
Here’s the uncomfortable truth about AI in creative work: generation has been commoditized.
Anyone with a laptop and twenty bucks a month can generate a logo, write marketing copy, produce a video script, create concept art. The output isn’t always great, but it’s often good enough. And “good enough” is now available at infinite scale.
This is the race-to-the-bottom fear that keeps creative professionals up at night. If everyone can generate, what’s left to sell?
But here’s what I’ve learned from two years of training creative professionals on AI tools: generation was never the valuable part. It was selection all along. The ability to look at ten options and know which one is right. The confidence to throw away the other nine. The capacity to articulate why.
That’s taste. And taste is the moat.
The “Creativity” Trap
When people talk about what AI can’t replace, they often land on “creativity.” It sounds good. It feels true. Creativity is human. Creativity is special. AI can generate, but it can’t create.
Except… that’s not quite right.
AI can now generate novel combinations at speed and scale that would have been impossible a decade ago. It can produce variations, explore style spaces, and propose directions that no human would have time to explore manually. Is that creativity? Depends who you ask. But it’s close enough to commoditize a lot of what used to pass for creative work.
What AI actually can’t do—yet—is evaluate. Judge. Decide. Choose.
AI doesn’t have preferences. It doesn’t have a worldview. It can’t tell you which of its outputs actually works for the context, the audience, the moment. It can only generate options and wait for you to pick.
That’s where the moat is. Not in the generating. In the selecting.
What Selectors Actually Do
At EA’s headquarters last September, I asked a room full of creative innovation executives: “What’s the one thing you do that you would never want to give up to AI?”
The answers were illuminating:
- “Knowing when something lands with an audience”
- “The moment when a piece finally clicks together”
- “Watching someone’s face when they see the work”
- “Making people laugh”
- “The judgment call that saves a project”
None of these are about generating. They’re about recognizing, evaluating, deciding. They’re about taste.
The selector doesn’t need to be the fastest generator. They need to be the most reliable judge. The one whose opinion you trust when there are a thousand options on the table and only one can ship.
Kevin Friel spent 25 years in Hollywood VFX. Dune, Detective Pikachu, Sonic, Bullet Train. The big shows. Now he teaches indie creators how to achieve blockbuster-quality production with AI tools. But when I asked him what AI can’t replace about his work, he didn’t talk about technical skills. He talked about knowing when a shot is “right”—the muscle memory of a quarter century spent making judgment calls under pressure.
“There’s something very punk rock about taking a tool that’s staring you in the face and using it to tell the stories that matter.”
Kevin Friel
That’s selector energy. The tool is a given. What you do with it is the differentiator.
The Ice Cream Test
Someone said to me recently: “The AI can’t eat ice cream for you. Only you can determine your preferences.”
It sounds trivial. But sit with it.
When AI can handle the calendaring, the accounting, the communications, the video editing… what actually remains?
- Preferences — what you want
- Judgments — what you think is good
- Evaluations — how you weigh tradeoffs
- Desires — what you’re trying to create
These are the stubbornly human things. Not the creating—machines can create. Not the executing—machines can execute. The wanting. The judging. The choosing.
Your taste is the accumulation of every judgment call you’ve ever made. Every time you said “yes, this” or “no, not that.” Every preference, stated or unconscious. Every standard you hold.
That’s what can’t be automated. That’s what can’t be averaged away. That’s your moat.
Building the Taste Document
I’ve started asking every creative professional I work with to build what I call a taste document. Not a style guide—though that’s part of it. A comprehensive record of their judgment patterns.
What goes in it:
Style Guide — How you sound. The rhythm of your sentences. The words you reach for. The words you never use. (I have an anti-glossary: “synergy,” “leverage,” “circle back,” “thought leader”—banned permanently.)
Worldview Document — What you believe about your craft, your industry, your audience. Not a manifesto. A working document that captures your actual positions so you can check AI output against them.
Glossary and Anti-Glossary — The thousand words you use most. The popular words you never use. The phrases that signal “this isn’t me.”
Decision Logs — Why you chose what you chose. Not just what shipped, but the reasoning behind it. The options you rejected and why.
These documents serve two purposes. First, they make AI output more useful—the AI has something to match against, something to aim toward. Second, and more important, creating them forces you to get clear about things you’d never articulated. The exercise of writing them is as valuable as having them.
Write for the Bot
There’s a video clip I show in every talk now. An anonymous OpenAI employee on Dwarkesh Patel’s podcast, voice disguised, identity hidden:
“If there are values you have which are not expressed yet in text, and if there are things that you like or want, if they aren’t reflected online, then to the AI, they basically don’t exist. And that is dangerously close to won’t exist.”
Anonymous OpenAI employee, Dwarkesh Patel podcast
Read that again.
If your taste isn’t documented, to AI it doesn’t exist. And as AI increasingly mediates what gets made, what gets seen, what gets funded—”doesn’t exist to AI” starts meaning “doesn’t exist.”
This is why the smartest creative professionals I know are writing. Not for audiences. For the bot. Getting their preferences, judgments, and values out of their heads and into text. Making their taste machine-readable.
Not because they want AI to replace them. Because they want AI to find them. To match with them. To amplify their specific perspective in a world drowning in generic slop.
The Selector’s Edge
Here’s what it looks like in practice.
Maya Bruck is a UX designer, formerly at Etsy. She came into our AI program as a “strategic skeptic”—interested but unconvinced. By the end, she could do on a one-hour phone call what used to take two weeks.
The magic wasn’t the tools. It was her judgment. Twenty years of UX decisions, distilled into the ability to look at AI-generated options and immediately know what would work. The AI generated the possibilities. Maya selected the right one. And Maya is now fast enough to be dangerous.
That’s the selector’s edge. You’re not competing with AI on generation speed. You’re competing on judgment quality. And if you’ve built up taste over decades of practice, you have a moat that no prompt engineer can replicate.
The Question
Close your eyes for a moment.
Think of one thing you do in your creative work that you would never give up to AI. Not because you couldn’t automate it—maybe you could—but because it’s the part that makes the work yours.
The judgment call you make when no one’s watching. The standard you hold that clients don’t know to ask for. The thing you fix at midnight because you can’t ship something that isn’t right.
That’s your taste. That’s your moat. That’s what’s worth building around.
Everyone’s talking about “prompt engineering” and “AI workflows” and “tool stacks.” Those matter. But they’re not the moat. The moat is taste. The moat is judgment. The moat is being the selector in a world of infinite generation.
Generation is cheap. Selection is not.
Your taste is your moat.
Going deeper: Make Culture, Not Content — A Field Manual for Creative Pros in the Synthetic Age, the longer companion to this piece.
Kris Krüg is Executive Director of BC + AI Ecosystem and teaches creative professionals how to apply AI through TheUpgrade.ai, which runs certification programs and custom training for designers, animators, VFX artists, and other creative professionals.
Discover more from Kris Krüg | Generative AI Tools & Techniques
Subscribe to get the latest posts sent to your email.