Building Taste and Judgement or Being Culture Fluid

The Two Things AI Can't Replace

If you have been using the latest AI models recently, you might feel another step-change in the progression. I certainly do.

In 2025, you might have had the same feeling I have had: looking at a piece of AI output and thinking, well, it's not perfect, it's...useable, you say hesitantly, but would I hand this to my boss? Hmm. A client? Hmm.

Those moments of hesitation were also moments of assurance. It told us that AI was still not good enough. But these days, AI has started to impress me again. And that makes me a little scared. Because it means the pace of change is about to accelerate.

Which brings us to the question we have been asking ourselves this whole time: what will humans be left with?

That is why I started investigating two concepts that I believe will define human value in the AI era: taste and judgement.

What Is Taste?

Recently, a meme went viral on the internet listing the traits of someone who "has taste."

No, you probably won’t have better “taste” if you start wearing Japanese denim and carry a Rimowa suitcase. We all recognize taste when we see it, but defining it is surprisingly hard. How do you train for taste? How do you get better at it?

Think of it this way. You ask AI to generate ten versions of your company's investor pitch deck. All ten are competent. They hit the right talking points, include the right data, follow a logical structure. But one of them has a narrative arc that will actually make a room full of VCs lean forward in their chairs. Knowing which one that is? That is taste.

Or consider a product manager reviewing the roadmap for the next quarter. The team has proposed twelve features. Engineering says they can ship eight. Every feature has data behind it: user requests, competitive analysis, revenue projections. But the PM looks at the list and knows that three of those features are the ones that will actually shift how customers feel about the product. The others are incremental. Being able to sense that difference, even when the spreadsheet says otherwise? That is taste.

Taste is discernment, the ability to evaluate quality, to look at a set of options and know which one belongs and which one does not. An editor who cuts the paragraph you loved most, and you realize the piece is better for it. A chef who resists adding one more ingredient because the dish is already complete. A founder who turns down a lucrative partnership because it would dilute the brand. In every case, the act is one of curation, not action. It is knowing what belongs.

AI can generate options at incredible speed. What it cannot do is feel the difference between something that is technically correct and something that resonates. That gap between "correct" and "right" is where taste lives.

Where Judgement Comes In

If taste is about discernment, judgement is about decision and action in context. Taste evaluates quality. Judgement reads a situation and decides what to do about it, right now, with imperfect information and real stakes.

Let me walk you through a scenario.

You work in sales. Your team is facing a major client proposal, and your job is to convince the CIO of a Fortune 500 company to sign a multi-million-dollar contract with your SaaS firm. On the team, there is a junior rep who was just hired, one year out of college. And then there is the ten-year veteran who is project-managing everything. One day, everyone hops on a Zoom call to pitch the client. In the room, you also have the Product Manager (person who built the product), the VP of Sales (should be a veteran sales guy), and maybe the CEO (the person who knows the company best). Whoever closes this deal will be a hero.

The junior rep starts going through the presentation. The client is not impressed. Everything he says sounds boilerplate, like it is rehearsed, or just lacking context.

Sensing they are losing the room, the ten-year veteran steps in. He notices the sweatshirt the client is wearing and does not even talk about the product. For the next ten minutes, they connect over the client's football team.

That is judgement. He read the room, assessed the situation in real time, and made a call: this client does not want a pitch right now; they want to know they are dealing with a real person. He broke the script because the moment demanded it. That ability to perceive what is happening, weigh it against experience, and act, all in the space of a few seconds, is what separates good judgement from just following the playbook.

Now consider a second scenario. New company, new client. The same group of senior guys come in with their usual cadence, but this time, it is not working. Their template has landed for years, so why not today? The junior rep steps in and mentions something that connects the client's business to an insight he picked up while interning in Japan. It creates a moment of genuine connection.

So judgement is not purely a function of seniority. It is about the richness of the mental models you bring to a live situation. The junior rep had a reference point the senior guys did not, and he had the presence of mind to deploy it at the right moment.

That is judgement in action.

It even recently happened to me. I was doing a keynote for a major German retailer. I thought I would be talking about one topic, but when I walked into the room and noticed the demographic, I picked another presentation topic to do instead. And it resonated with them so much more. Replace me with a robot, and it might have just gone with the pre-packaged keynote topic.

Here is the key distinction: taste could have told you which pitch deck to use before the meeting started. Judgement is what tells you to throw the deck away mid-meeting because the room has shifted. Taste is pre-game. Judgement is real-time.

AI can draft a contract, flag risks in a financial model, or generate a dozen strategic options for a product launch. But someone has to decide which risks to accept, which strategy to pursue, and when to override the data because something feels off. AI gives you options. Judgement picks the one that matters.

How to Systematically Build Better Taste and Judgement

If taste and judgement are what separate us from AI, the natural follow-up question is: can we actually train for them? Or are they just something you either have or you don't?

The research says yes, you can train for them. But not in the way most people think. You do not build taste by reading a book about taste. You build it through a specific kind of exposure, feedback, and reflection. The science behind expertise and intuitive decision-making gives us a surprisingly clear roadmap.

Start with Being Culture Fluid

A few years ago, I coined a term in my TEDx talk that I think captures the foundational skill underneath taste and judgement: Culture Fluid. Being Culture Fluid means getting comfortable with the uncomfortable. It is the ability to move between cultures, contexts, industries, and generations without losing yourself, but also without rigidly holding onto a single frame of reference. It is about the willingness to step outside your default worldview and genuinely absorb how other people think, work, and make decisions.

I developed this concept originally in the context of bridging Eastern and Western business cultures, something I navigated daily during my years at Alibaba. But as I have watched AI reshape the professional landscape, I have realized that being Culture Fluid is not just a DEI principle. It is the operating system for building taste and judgement in the AI era.

Here is why. Every research-backed method for developing expertise, from pattern recognition to deliberate practice to feedback loops, depends on one prerequisite: the willingness to expose yourself to things that are unfamiliar and sit with them long enough to learn. That is Culture Fluid in a sentence. The junior rep who connected with the client through his Japan internship? He was Culture Fluid. He had put himself in an unfamiliar context, absorbed its patterns, and carried them forward into a moment where no one else in the room had that reference point.

You cannot build a diverse pattern library if you only consume within your own bubble. You cannot develop real-time situational judgement if every room you walk into looks and sounds the same. Being Culture Fluid is what opens the door to everything that follows.

Build a Bigger Pattern Library

Taste, at a cognitive level, is pattern recognition applied to quality. The more high-quality and varied examples you have absorbed, the better your internal filter becomes. This is why people with great taste tend to be voracious consumers across disciplines. They read fiction and financial reports. They attend art exhibitions and engineering conferences. They study how a great restaurant menu is structured and how a compelling legal brief builds its argument.

Each exposure deposits a new reference point. And when you later encounter something, whether it is a pitch deck, a product design, or a piece of writing, your brain is unconsciously comparing it against that library. The richer the library, the sharper the filter.

But here is the question I get asked most often when I talk about this: Does the exposure need to be random? Or is there a smarter way to curate it?

The answer, drawn from several decades of research on creativity and cross-domain thinking, is that it should be neither purely random nor rigidly planned. The sweet spot is what I would call intentional intersection-seeking.

Frans Johansson captured this idea in his book The Medici Effect, named after the fifteenth-century Italian banking family whose patronage of artists, scientists, philosophers, and architects in Florence helped ignite the Renaissance. Johansson's central argument is that the most powerful innovations happen at the "Intersection," the place where concepts from different fields, disciplines, and cultures collide. The Medici family did not fund random work. They brought together people from diverse, high-quality disciplines and gave them proximity to one another. The breakthroughs came from the combinations.

David Epstein builds on this in Range: Why Generalists Triumph in a Specialized World. His research shows that in complex, unpredictable environments (which describes most modern professional work), generalists consistently outperform narrow specialists. The key mechanism is what Epstein calls deep analogical thinking: the ability to recognize structural similarities between problems in completely different fields. The best problem-solvers do not just draw from their own domain. They borrow frameworks from distant ones. An architect's concept of load distribution might reshape how you think about team structure. A documentary filmmaker's approach to narrative pacing might change how you design a product onboarding flow.

So what does this mean practically? Not everything needs to be random, and not everything needs to be adjacent. The formula looks something like this:

One: Anchor in your domain. You need depth in your own field to have something to connect to. Taste without expertise is just preference. The PM who can sense which features matter has that instinct because she deeply understands her product, her users, and her market. That depth is the foundation.

Two: Seek distant domains that share deep structural problems. Instead of consuming randomly, ask yourself: what other fields are solving a version of the same problem I face? If you are in enterprise sales, do not just study other sales methodologies. Study how hostage negotiators build trust under pressure. Study how documentary filmmakers earn a subject's confidence in the first ten minutes. Study how great teachers hold attention in a room where nobody chose to be there. These are structurally similar challenges in wildly different contexts, and the solutions transfer in ways that are not obvious.

Three: Leave room for genuine randomness. Not every input needs to be strategic. Some of the most valuable patterns in your library will come from places you never planned to look. The magazine you picked up at the airport. The conversation with a stranger at a dinner party. The novel that had nothing to do with your work but gave you a metaphor you have used a dozen times since. Johansson specifically recommends "intersection hunting": deliberately picking up things with no apparent connection to your current problem and seeing where they lead. Structured randomness, paradoxically, is one of the most productive creative strategies.

This is where being Culture Fluid becomes a practical advantage, not just a mindset. The willingness to cross boundaries, to sit with unfamiliar contexts, to absorb how different cultures and industries think, is what fills your pattern library with the kind of diverse, high-quality inputs that produce exceptional taste. Curation gives you depth. Randomness gives you surprise. The intersection of the two is where taste sharpens fastest.

--

Hello! I’m Sharon Gai, author of How to Do More with Less: Future-Proofing in an AI-Driven World  and a keynote speaker on AI and its effects on workers and the future of work.