Taste Is Technical
Ask a chef why they added lemon at the end and they'll say it "needed it." Ask a designer why they chose that typeface and they'll say it "felt right." Ask a filmmaker why they cut there and they'll say "instinct."
This is what people point to when they say AI will never have taste. That taste is something ineffable. Something human. The last fortress.
They're wrong. Taste is technical. It's just technical in a way the maker has forgotten is technical.
The chef added acid because the dish was flat. There's a reason. Fat needs brightness. Sweetness needs contrast. These are rules. The chef learned them, practiced them for years, and eventually stopped being able to see them as rules. They became "feel."
That's not mystery. That's mastery so deep it looks like magic.
Every domain has this. A typographer picks a font because of its x-height relative to the line length, the counter shapes, the historical context of the project. They don't think about any of this consciously. They just "know." But if you sat them down and forced them to explain, they could. It would take hours. The explanation would be boring and granular and entirely mechanical.
The reason people think taste is beyond computation is because the people who have it can't articulate it quickly. The knowledge is tacit. It lives in the body, in the hands, in thousands of micro-decisions made so fast they feel like one big intuition.
But tacit doesn't mean mystical. It means compressed. A neural network that's been trained on enough examples also develops "taste." It doesn't call it that. It calls it weights. Same thing.
Here's what's actually happening when someone defends taste as uniquely human: they're confusing the feeling of a skill with the structure of a skill. It feels like intuition. It feels like soul. But underneath, it's a decision tree so vast and so practiced that the person running it forgot it was a decision tree.
This matters because the people clinging to taste as the last human advantage are building their identity on a misunderstanding. Taste isn't the thing that can't be automated. Taste is the thing that's hardest to automate. Those are very different statements.
Hard means it takes more data, more nuance, more context. It doesn't mean impossible. And "more" is exactly the direction AI is headed.