Special Report: Taste in the Age of AI
Sam Parr just told 1.5 million people that taste is a moat. The New Yorker, New York Times, and Silicon Valley also weighed in this week. Here's what they're missing.
This isn’t your regular Saturday newsletter. I’m sending this mid-week because something happened this week that I can’t stop thinking about, and it sits at the exact intersection of everything we talk about here.
Silicon Valley’s New Favorite Word
Paul Graham posted on X: “In the AI age, taste will become even more important.”
The founder of the AI design tool Framer said “great taste” is what will create the best new products. Marc Andreessen suggested that once AI handles execution, picking what to build (venture capital, essentially) “may be one of the last remaining fields.” A startup founder wrote that “personal taste is the moat.”
The New Yorker’s Kyle Chayka is skeptical of this. He calls it “taste-washing.” I think he’s half right. The tech companies aren’t wrong that taste matters. They’re just defining it too narrowly. They’re talking about taste as curation, as selection, as knowing what to build. That’s real. But it’s the surface layer of something deeper.
The tech definition of taste is clear: the ability to choose what will be profitable. Taste as strategic decision-making. Taste as competitive advantage. Taste as moat.
I’ve spent over a decade helping people develop taste, and I think they’re pointing at something real. They’re just starting halfway through the process.
What Taste Actually Is
Taste, in the way I use it, is knowing your own values clearly enough to make decisions from them. It’s the process of figuring out what actually matters to you, not what you’ve been told should matter, and choosing accordingly.
Perception is part of it. You have to notice what resonates. But the real work is connecting that resonance to something deeper: who you are and what you value.
AI is excellent at the parts that used to take the most time. It can scan a wine list, match food pairings, surface options you’d never find on your own. That’s genuinely useful. What it can’t do is tell you what you value. It can optimize for your past preferences, but it can’t help you discover that your preferences were borrowed in the first place.
I’m currently auditing the Wine Scholar Guild’s Tasting Diploma, and their entire curriculum is built on this distinction. They teach tasting as an embodied act: start with your gut reaction, your physical response, how the wine makes you feel. Analysis comes later. The tasting toolkit they’ve developed explicitly reverses the analytical grid approach that dominates wine education, because grids treat the taster like a machine processing inputs.
The interesting thing is that this approach and the tech conversation are pointing at the same truth from different directions.
Silicon Valley is discovering that taste creates value. Wine educators are discovering that taste starts with self-knowledge. Both are right. The gap is that most people skip the self-knowledge part and jump straight to curation.
The NYT Angle: AI at the Table
The New York Times reported this week that sommeliers are watching diners pull out ChatGPT to photograph wine lists and get recommendations at the table.
One guest described wine lists as “a quiz you didn’t prepare for.” I heard versions of that sentence for years as a sommelier. The tight smile, the quick scan, the nervous “what do you recommend?”
AI solves that anxiety. You get a decent recommendation. You avoid embarrassment. The evening moves on.
But here’s what bothers me about that solution: you go from deferring to a critic to deferring to an algorithm. The posture hasn’t changed. Someone (or something) else is still making the decision for you.
A sommelier reading a table on a Friday night does something AI cannot replicate. They’re deciding whether someone wants to learn or just wants to feel taken care of. They’re noticing the couple who hasn’t made eye contact in twenty minutes and choosing a bottle that might change the energy. They’re building a relationship with a guest that makes them come back.
AI is an extraordinary recommendation engine. Fast, comprehensive, getting better every month. It serves a purpose, but the value a great sommelier brings to a dining room lives in everything the shortcut can’t capture.
The Podcast That Proves the Point
Four days ago, Sam Parr released an episode of My First Million called “Taste is your moat.” He told his 1.5 million listeners that taste is “probably the most important thing you can learn about right now.” He laid out a four-step process: decide what you want to say, blindly copy people you admire, learn the rules underneath what they’re doing, then study history.
He used fashion as his example. He unfollowed everyone on Instagram, bookmarked people who “looked cool,” bought what they wore, read books on the rules of menswear, traced the history of workwear and Ivy style back to its roots. He discovered that the styles he was drawn to reflected his values: Midwestern stoicism, frontier adventure, generational tradition.
He also traced Dr. Dre back through George Clinton, Motown, and gospel to show how musical taste works the same way: learn the tradition, master the rules, then break them.
The process is solid. And something he said at the end stuck with me: “Being around beautiful stuff, stuff that sings to you, it honestly makes my life happier. And a lot of times I always felt like I knew that, but I didn’t have the language to describe what I liked or why I liked it.”
That’s real. That’s the feeling I hear from readers every week.
But here’s what Sam’s process doesn’t include: perception. His framework starts with “decide what you want to say.” Mine starts with “notice what resonates.” Those are fundamentally different entry points.
Sam’s process is about projection: how do I communicate my identity outward? The work we do here is about reception: what am I actually experiencing, and can I trust my response to it?
Both matter. But in a world where AI can handle projection (it can build your brand, design your website, curate your aesthetic), the skill that remains irreplaceable is the one Sam skipped. The ability to perceive. To use your values as a lens. To feel the weight of a wine on your tongue and know whether that weight is comforting or excessive.
Sam correctly points out that there’s a process to building taste. He’s cluing millions of people in to something important. I just think the process has a step zero that most people skip: learning to pay attention to your own experience and identify your values before you decide what to project.
The Missing Step
Paul Graham is right; if you know what good looks like, you make better decisions, build better products, live a more intentional life. Sam Parr is right too. There’s a learnable process: study what you admire, copy it, learn the rules, understand the history.
But both of them are describing what happens after you’ve already done the foundational work. The part that gets skipped is the part before the process: figuring out what you actually value.
The 50% of New York Times readers who preferred AI-generated prose over Hilary Mantel aren’t lacking taste. They’re living in an environment that has trained them to optimize for speed, familiarity, and ease. As Chayka writes, the online ecosystem has “become so polluted, so fragmented, deceptive, overstimulating, ersatz, that it has warped our ability to exercise taste at all.”
Wine is a useful antidote to that. Not wine as status symbol or Instagram prop. Wine as a practice of slowing down long enough to notice what actually resonates with you. A glass that forces you to pay attention, respond, and decide for yourself.
The Real Moat
The real moat is self-knowledge. Knowing what you value clearly enough to make decisions from those values, even when the algorithm, the critic, or the confident person next to you suggests something different.
AI makes this easier in some ways. It removes the friction of research, surfaces options you’d never find, handles the logistics. That’s genuinely valuable. What it can’t do is tell you what matters to you. It can optimize for your past behavior, but it can’t help you notice that your past behavior was borrowed.
That’s the work. Not fighting AI. Not resisting the tech conversation about taste. Just doing the slower thing underneath it: building your own internal compass through attention, exposure, and honest self-assessment.
Paul Graham says taste is the moat. I agree; I just think most people need help building it. That’s what this newsletter is for.
Go Deeper
The sources behind this issue, if you want to follow the threads:
“Why Tech Bros Are Now Obsessed with Taste” (The New Yorker, March 2026)
“Can A.I. Give Better Wine Advice Than a Sommelier?” (The New York Times, March 19, 2026)
“How Do Wine Programs, Sommeliers, and AI Interact?” (VinePair Podcast, March 23, 2026)
“Who’s a Better Writer: A.I. or Humans?” (The New York Times interactive quiz, March 9, 2026)
Sam Parr, “Taste Is Your Moat” (My First Million, Episode 809, March 27, 2026)