Taste, Nuance, Creativity, and the Thing AI Can’t Manufacture

I’ve been working on a framework called the Human Capability Model™ for a while now. Long enough that I sometimes forget not everyone has been living inside it with me. One of the concepts at the center of it is something I call the Human Continuity Barrier, which is, in the simplest terms I can offer, the ever-shifting line that marks where humans can go and AI cannot.

I say ever-shifting because it moves. Constantly. What was firmly on the human side of that line two years ago is now fully automatable. The line will keep moving. That’s not a crisis, it’s just physics.

But here’s what I keep coming back to: there are things on the human side of that barrier that I genuinely don’t know if AI will ever cross. Not because I’m being romantic about it. Because I can’t figure out the mechanism by which it would.

Then I read something this week that felt like a quiet confirmation of exactly that.

Chinese tech companies, the most AI-forward organizations on the planet right now, are bypassing universities and recruiting high schoolers directly. Not for technical skills. For creativity. The belief is that younger candidates may be more likely to imagine things that don’t yet exist.

The companies building the most sophisticated AI systems in the world are going out of their way to find humans who can imagine things that don’t exist yet. They are not, notably, prompting their own models for this.

That’s not irony. That’s a data point.

The companies building the most advanced AI systems in the world are recruiting for human creativity—because their own models can’t provide it.

The stack that AI can’t replicate

When I think about what sits on the permanently (or at least stubbornly) human side of the Human Continuity Barrier, I keep landing on the same three things: taste, nuance, and creativity. And more importantly, I keep landing on what produces them.

Experience. Specifically, the kind where you fall flat on your face.

Taste is the ability to look at something technically correct and know it’s wrong. To look at something rough and know it’s right. That calibration doesn’t come from consuming enough examples. It comes from having been wrong in ways that cost you something. From having sent the polished version and watched it land flat. From having gone with your gut on the messy version and watched it connect in ways you couldn’t have predicted or explained. You don’t develop taste by studying taste. You develop it by falling on your face enough times that your instincts start to know things your brain hasn’t caught up to yet.

AI can produce technically correct outputs all day. What it cannot do is feel the difference between correct and true. It has never fallen on its face. It has never had to.

Nuance is knowing which version of the truth to say, to whom, at what moment. Not just the right message for the right audience (AI can approximate that). Nuance is the specific word that carries weight with this particular person in this particular context, on this particular day. That comes from having watched the right words land badly at the wrong moment. From having delivered feedback that was accurate and still left the room worse than you found it. From falling on your face in front of people whose opinion mattered and having to figure out what went wrong.

Creativity, real creativity, is not recombination. It’s construction from something lived. Nassim Taleb’s argument in Skin in the Game is relevant here: people who bear no consequence for being wrong give you a fundamentally different, and worse, kind of judgment than people who do. AI has no skin in the game. It cannot be embarrassed, fired, or proven wrong in a way it has to sit with overnight. It has never had to pick itself up off the floor and recalibrate. It has patterns from people who have done all of those things. That is not the same as having done them.

Creativity without scar tissue is just novelty. It can be interesting. It’s rarely true.

What this looks like in practice

I was walking through the Human Capability Model™ recently with Chris Long and Mike Bollinger, who heads up thought leadership at Cornerstone OnDemand. We got into the Human Continuity Barrier specifically, and Mike pushed back on the word “barrier.” Too static, he said. Implies an endpoint. “The notion of a barrier implies some sort of endpoint, and what you’re really trying to describe is something that has a degree of fluidity.”

He was right. The pushback made the definition sharper. It’s not a wall. It’s a membrane, permeable from the human side because we’re the ones building the tools. AI hasn’t figured out how to push through from its direction yet. Maybe it will. But the reason it hasn’t is the same reason you can’t shortcut the stack above: taste, nuance, and creativity are outputs. Falling on your face, repeatedly, consequentially, is the only input that generates them reliably.

Mike also said something that stuck: “Right now I think the big issue is that people don’t know what they’re capable of. And they’re rapidly being relegated to being a cog in the wheel.”

That’s the actual problem. Not that AI is too capable. That humans are underselling what they bring to the line.

Taste, nuance, and creativity aren’t soft skills. They’re the hardest ones we have—and they’re still exclusively human.

What this means for the humans in the room

The Human Capability Model™ maps eight capability vectors: Strategic Navigation, Architect and Systems Design, Capability Discovery and Matching, Execution and Delivery, Augmentation and Automation Integration, Learning and Adaptation, Coordination, Culture and Human Infrastructure, and Innovation, Exploration and Nonlinear Value Creation. Every one of these existed, in some form, before there was a computer in the building. They map to things humans have always done.

The Human Continuity Barrier sits around all of it. And right now it’s getting a lot of attention because the AI conversation is loud and the displacement fear is real. But the barrier isn’t new. What’s new is how quickly the line is moving, and how little time most organizations are spending thinking about what that means for the humans they employ.

The China story isn’t about AI replacing people. It’s about the most technically sophisticated operators in the world making a deliberate bet on the one thing their systems can’t produce: a human who hasn’t been told yet what’s impossible, and who has enough scar tissue to know the difference between an idea and a good one.

That’s the capability they’re recruiting for. And it’s only available in humans. (For now. I reserve the right to be wrong about this.)

The bar for human work just got higher. Not because AI is smarter. Because the things that now require a human are the things that always required the most from us. Judgment. Conviction. The ability to fall on your face, get back up with new data, and try again knowing it might happen twice more before you get it right.

Those aren’t soft skills. They’re the hardest ones we have.

And they’re still ours.