Confident Mediocrity: On AI, taste, and what happens when the machine works faster than your judgment.

One of the more useful Substacks I read is from Greg Shove and Taylor Malmsheimer from SectionAI, the online school that has shaped so much of my way of thinking about AI, not as it’s applied to creativity, but how it’s applied to business. I completed my Mini-MBA in AI for Business with Section a while ago.¹ In the most recent newsletter, Greg and Taylor point to two observations. They weren’t connected, rather part of a long list of observations.

I think they should be.

The first: the work ratio has flipped. For most of modern professional history, work looked roughly like this — 10% deciding what to make, 80% making it, 10% checking whether it was any good. That middle 80% was the job. That’s where the hours went, where the craft lived, where you proved your value by volume of output.

Now it’s closer to 40/20/40.²

Your numbers may look different right now. But start asking your team how they’re actually allocating their time on any given project, and watch what you find. The shift is already happening underneath the official version of how work gets done.

The making — the part machines have swallowed — is down to 20%. The bookends have doubled. Human time is now concentrated at the front, where you decide what matters and instruct the machine on how to pursue it, and at the back, where you judge what came out and refine it with something the machine doesn’t have.

We’ll get to what that something is.

The second observation: you feel like a bottleneck, yet more valuable at the same time. Machines work faster than you. You are simultaneously the most important part of the process and the thing slowing it down. That tension is uncomfortable.

It’s also correct.

Most people I talk to are experiencing this and don’t have language for it. They know something has shifted. They feel both more necessary and more exposed. They’re producing more, faster, and somehow feel less confident about what they’re contributing.

That’s not an accident. That’s the ratio doing its work on you.

Taste Is the Job Now

Here’s what’s actually happening at both bookends.

At the front, when you decide which tasks matter and how to instruct the machine to pursue them, you are making aesthetic and strategic judgments in rapid succession. You’re not filling out a brief. You’re transmitting a point of view. The quality of what the machine produces is upstream of the prompt, and the quality of the prompt is upstream of your accumulated sense of what good looks like.

At the back, when you’re judging output and refining it, you’re doing the same thing. You’re not proofreading. You’re tasting.

So what the ratio shift has actually done — beneath all the productivity language, beneath the frameworks and the efficiency gains — is move taste to the center of professional value.

Taste. That unfashionable word. The thing that can’t be credentialed, can’t be templated, can’t be prompted into existence. The thing that senior people used to carry quietly and organizations systematically undervalued because it was hard to measure and easy to dismiss as subjectivity.

AI didn’t create this. It exposed it.³

For decades, organizations optimized for the middle 80%. They built processes, workflows, approval chains, and output metrics all designed to manage the making. Talented people learned to survive — and thrive — by being excellent producers. Fast. Reliable. High volume. The judgment at the bookends was often perfunctory, a formality before the next production cycle began.

Now the making is nearly free. And suddenly everyone is being asked to do the thing their entire career never trained them for.

To have a point of view. To know what good looks like. To be the judge.

Most people weren’t ready for that. Not because they lack intelligence. Because taste is built through years of active, curious, opinionated engagement with the world — and the professional systems most of us moved through actively discouraged that kind of engagement in favor of execution.

This is why so many AI outputs feel like almost-good. Not because the models aren’t capable. Because the humans directing them haven’t yet developed — or rediscovered — the judgment required to push past almost.

A Thousand Images to Find Thirty

I finished a project this week that made this concrete for me. A client needed thirty to thirty-five images for a website rebrand — their visual language around a specific strategic idea that I’ll call talent as infrastructure.

Over two days, I generated close to a thousand images. Architecture as a literal symbol. Architecture as metaphor. The integration of people and structure. Signature color. Expression. Open space.

I knew what I was looking for before I started. The machine didn’t.

It produced prolifically and without judgment, and I moved through the output the way you move through contact sheets — looking for the frame that justifies the entire roll. We landed thirty-plus images that were exactly right. But we also landed something else: a recipe. A prompt architecture that encoded everything the process had taught us about what good looked like for this client, this idea, this visual world.

The taste came first. The machine made it scalable.⁴

The Bottleneck Is the Point

Back to that second observation. You feel like a bottleneck.

You are.

The machine has no patience for your deliberation. It doesn’t care that you need to sit with something before you know whether it’s right. It doesn’t understand that the pause before you say “this isn’t quite it” is not inefficiency — it’s the whole game. It will produce the next version before you’ve finished evaluating the last one. It will produce ten versions, cheerfully, and wait.

That gap — between machine speed and human judgment — is where most people are losing their minds right now. They interpret it as falling behind. As inadequacy. As evidence that they are, in some fundamental way, not built for this moment.

They have it exactly backwards. The friction is the feature.⁵

Without human judgment in the loop, you don’t get efficiency. You get confident mediocrity, produced at inhuman speed, distributed at inhuman scale. You get slop that doesn’t know it’s slop, wearing the costume of quality because it’s grammatically correct and structurally coherent and utterly, forgettably fine.

The bottleneck — you, your taste, your refusal to sign off on almost-good — is the only thing standing between your organization and that outcome.

This is genuinely new territory for most professionals. Not because the skills required are exotic, but because the source of value has moved somewhere most people weren’t watching. It moved out of execution and into judgment. Out of production and into discernment. Out of the comfortable middle — where effort was visible and progress was measurable — and into the bookends, where the work is harder to see and harder to defend but more consequential than anything that came before it.

You are not slowing the machine down. You are the reason it produces anything worth keeping.⁶

Your Point of View Is Now Your Job Description

So here’s what I’d ask you to sit with.

The professional skills that got you here — speed, output, reliability, volume — are not worthless. But they are no longer the point of differentiation. The machine has absorbed them. What remains, what the machine cannot absorb, is the thing you may have spent your entire career treating as secondary.

Your point of view.

Not your credentials. Not your process. Not your frameworks or your methodologies or your years of experience catalogued on a LinkedIn profile. Your actual, specific, hard-won, opinionated sense of what good looks like and why it matters.

That’s what’s being asked of you now. At the front of every task, when you decide what’s worth doing and how to pursue it. At the back, when you look at what the machine returned and decide whether it’s finished or whether it needs you again.

The ratio shifted. The job changed. And the people who will navigate this moment best are not the ones who figure out how to use AI faster. They’re the ones who finally take their own taste seriously.

Taste isn’t a personality trait.

It’s a practice. It’s built by consuming voraciously and reacting honestly — not just noting that something works, but demanding of yourself an answer to why. It’s developed by having opinions and testing them, by saying “that’s not quite it” enough times that you start to know what right actually feels like. It’s sharpened by exposure — to great work, to failed work, to work that almost made it and didn’t.

Most professionals stopped doing this somewhere in their thirties when the calendar filled up and consumption became passive. The good news is it’s recoverable. The bad news is there’s no shortcut. You build taste the same way you always did — by paying fierce, stubborn, opinionated attention to the world around you. AI just made having it worth more than it’s ever been.

But while you’re building it, borrow it.⁷

Find the people in your orbit — inside your organization or outside it — whose taste you trust. People with visible histories of knowing what good looks like, and a track record of saying so. Work next to them. Watch how they evaluate. Ask them not what they think of something, but why.

The accumulation of taste takes time. The acceleration of it doesn’t have to. Surround your team with people who already have it, and let that standard become the new baseline. Individual taste compounds into organizational judgment. And right now, organizational judgment is the scarcest resource in the building.

Start there.

Christopher Smith believes that storytelling is a system, taste is a practice, and most organizations are producing confident mediocrity at scale without knowing it. He helps them do better. His work spans strategy, creative direction, and AI — with clients across Asia and beyond.