Quality & taste

I’ve been using large language models (LLMs) as a bicycle for my mind: learning at warp speed, writing in new languages, exploring wild ideas, building things faster than ever. It still feels like magic. But it’s also strange and unsettling, because the deeper I go, the more I notice something missing.

These models don’t know what good is. They don’t know what matters. They don’t know how to care. And that’s the real gap.

LLMs don’t have taste. Not the way humans do. They don’t feel the difference between something that works and something that’s heartfelt. They have no opinions, hesitations, convictions, or hills to die on. They’ll give you a perfect sentence and then offer its opposite with equal confidence. They’re strange tools, full of ghosts.

We often forget that quality is something we feel. It’s not a checklist or a score. It’s not just clean code, balanced design, or a thoughful abstraction. It’s the signal that someone cared. That they paid attention to the details without losing sight of the whole. When something feels truly good, we’re not just seeing polish, we’re sensing a human heart behind the work. A point of view. A set of values, made tangible.

Taste sharpens that point. It’s the inner compass you build over time, through trial and error, through regret and surprise. Taste isn’t just about what you like. It’s what you notice. What you reach for and what you reject. You don’t always know why something feels off, but you know it does. That’s taste doing its job.

LLMs don’t have that. They can only remix the past. They don’t know why one thing bangs while another clangs. They don’t care if your product feels clumsy or if your life falls apart. They just generate. Endlessly.

And that’s the difference.

Humans feel their way through things. We chase hunches. We overthink. We throw work away because it doesn’t feel right. And that’s not inefficiency, that’s the whole point. That’s how we find the thing with soul, with spark, with weight.

LLMs can help us work faster. But they can’t tell us where to go. They can’t say what’s worth making. That’s still on our meat-computers to figure out.

Because at the core of all this is one simple, stubborn fact:

We care.

And for now, caring is something only humans can do.