Here's a question I've been wrestling with: When AI can do everything, what's left for humans to do?

The standard answers—creativity, emotional intelligence, physical work—are all vulnerable. AI is already creative. Emotional intelligence is being simulated. Physical work is being automated. Each "human only" domain keeps shrinking.

But there's one thing that I don't think AI can replicate: taste.

The Execution Explosion

We're living through an execution explosion. Things that used to take teams and months now take individuals and days. Code, design, writing, music, video—all becoming trivially producible.

This is usually framed as democratization. Anyone can make anything. True. But what's less discussed is the flip side: when anyone can make anything, most of what gets made is garbage.

Not because the tools are bad. The tools are incredible. But because deciding what to make is harder than making it.

The world is about to be flooded with technically competent, aesthetically coherent, completely mediocre content. AI makes it easy to hit the quality floor. But hitting the quality ceiling—making something that actually matters—that's getting harder, not easier.

What Is Taste, Really?

Taste isn't preference. Preference is liking chocolate over vanilla. That's personal and arbitrary.

Taste is the ability to discern quality. To know that this is better than that, and to be able to articulate why. To recognize when something works and when it doesn't, even when you can't immediately explain the difference.

More importantly, taste is about knowing what's worth doing. Not just what's possible, but what matters. What will resonate. What will last.

This is fundamentally different from execution. Execution is about how. Taste is about what and why.

The Curation Problem

AI is excellent at generation. You can ask for a thousand variations of anything and get them instantly. But which variation is the right one?

This is the curation problem. AI gives you options. It doesn't tell you which option to choose. It can generate a thousand logos, but it can't tell you which one will define your brand for the next decade.

Yes, AI can help with curation too—you can ask it to rank the options, explain the tradeoffs, predict which will perform better. But these are just more generations. At some point, a human has to decide. And that decision is taste.

The person who can look at a thousand AI-generated options and pick the right one—or better yet, know what to ask for in the first place—that person is more valuable than ever.

Taste Is Accumulated Judgment

Here's what makes taste hard to automate: it's not a skill you can learn from a dataset.

Taste is accumulated judgment from lived experience. It's knowing what works because you've seen thousands of things work and fail. It's pattern recognition trained on a lifetime of cultural immersion, emotional responses, and contextual understanding.

AI can simulate this to some degree. It can analyze what has worked historically and predict what might work in the future. But taste isn't just prediction—it's judgment. It's knowing when to break the rules. When to do something that hasn't been done. When to trust your gut over the data.

The best creative work often violates patterns. It succeeds because it's unexpected. AI trained on what worked before will always be biased toward what's already been done. Taste allows you to see what should be done.

The Taste Hierarchy

I think we're going to see a new hierarchy emerge:

At the bottom: People who can't execute and don't have taste. They've always been at the bottom, and they'll stay there.

Rising slightly: People who could execute but couldn't get taste. They used to be limited by their skills. Now AI gives them execution ability. They'll be able to produce more, but they still won't know what to produce.

Falling significantly: People who could execute but didn't need taste. Specialists who did technical work without making creative decisions. Their execution skills are being commoditized.

At the top: People with taste. Whether or not they could ever execute is now irrelevant—they can direct AI to execute. What matters is that they know what's worth making.

This is a profound inversion. We've spent decades building education systems that reward technical execution. Following instructions. Solving defined problems. Producing correct outputs.

None of that develops taste.

How Taste Gets Developed

If taste is about to become the most valuable skill, how do you get it?

The uncomfortable answer: I'm not sure you can develop taste intentionally. You can expose yourself to conditions that might produce it, but there's no guaranteed path.

What seems to help:

Consuming widely and deeply. People with taste have usually consumed enormous amounts of their domain. Not just the hits—the misses too. Understanding why something failed is as important as understanding why something succeeded.

Creating prolifically. Making a lot of things, even bad things, develops judgment. You learn to recognize quality by producing quantity.

Getting feedback from reality. Taste developed in isolation isn't taste—it's preference. Real taste comes from seeing how your judgments play out in the world. What you thought was good that failed. What you thought was bad that succeeded.

Diverse experience. The best taste often comes from cross-domain exposure. Seeing how quality manifests in different fields helps you recognize universal principles.

Time. There's no shortcut. Taste takes years to develop. This is why it's hard to automate—it requires living, not just learning.

The Taste Economy

What does an economy look like when taste is the primary valuable skill?

First, I think we'll see a massive premium on curators. People who can filter the noise and surface quality. This already exists—editors, critics, tastemakers—but their importance is about to explode.

Second, personal brands become everything. If your value is your judgment, then trust in that judgment is your most valuable asset. Being known for good taste becomes more valuable than having any particular skill.

Third, the long tail gets longer and thinner. AI makes it easy to produce, so production explodes. But attention is finite. The gap between what succeeds and what doesn't will become more extreme.

Fourth, traditional credentials lose value. A degree shows you can execute predetermined curricula. It says nothing about taste. I expect informal signals of taste—portfolios, track records, reputations—to replace formal credentials.

What This Means for You

If you're early in your career: prioritize taste development over skill development. Yes, you need basic competence. But above a threshold, more skills won't help you. Better taste will.

If you're mid-career: ask yourself whether your value comes from execution or judgment. If it's primarily execution, you're in trouble. Start deliberately developing taste now.

If you're building a company: hire for taste, not just skill. The skills are becoming automated. The taste isn't.

If you're investing: look for founders with demonstrable taste. The ability to pick the right thing to build is more important than the ability to build it.

The Last Human Advantage

I titled this "the last human advantage" intentionally. I'm not certain taste will remain human-only forever. Maybe AI will eventually develop genuine judgment, not just simulated judgment.

But I think taste is the last thing to fall, if it falls at all. It requires something that AI currently lacks: the experience of being human. Of caring about things. Of living in the world rather than modeling it.

Taste emerges from the intersection of knowledge, experience, and values. AI can have knowledge. It might simulate experience. But values—genuine values, not programmed preferences—might be uniquely human.

If that's true, then taste is our last advantage. We should start treating it that way.

The question isn't whether you have taste. Everyone thinks they do. The question is whether you're deliberately developing it, or just hoping it develops on its own. In the coming economy, that distinction will matter more than any skill you could learn.