When Silicon Valley Talks 'Taste,' What's It Really Hiding?

Opening

Reader, do you know what the hottest keyword in Silicon Valley has been lately? It’s not AI, and it’s not agents. It’s “taste.” I wrote about this in my book last year, and I touched on it in a blog post 2 years ago too. But lately I’ve had a strange thought. At some point, everyone started saying the same thing. And my reaction wasn’t “See, I was right all along” — it was “Wait, is this actually right?” Because when everyone starts saying the same thing, that’s usually a sign the market has entered its maturity phase. Which also means it’s time to look for something new. (I’ll unpack that chasm theory another time.)

So I looked around, and sure enough, Y Combinator co-founder Paul Graham had posted this on X: “In the age of AI, taste matters more. When anyone can build anything, differentiation comes from choosing what to build.” Cloudflare CTO Dane Knecht chimed in, saying “In 2026, taste becomes the differentiator in engineering,” and OpenAI’s Greg Brockman joined the chorus too. Koen Bok, founder of the AI design tool Framer, declared on a podcast that “great taste makes the best products.”

The New Yorker’s tech-culture columnist Kyle Chayka took this phenomenon head-on. He calls it “taste-washing” — dressing anti-human technology in humanist clothing.

Which raises a question: is the “taste” Silicon Valley is talking about the same taste we’ve always known?

Why “Taste,” Suddenly — The Paradox of a World Where Production Is Democratized

Let’s start with why this discourse exploded in the first place. The core reason is simple: AI has nearly erased the barrier to entry for “making things.”

With coding assistants like Anthropic’s Claude Code, someone with zero programming experience can build an app. Image-generation AI lets anyone produce visuals, and LLMs¹ let anyone write. Technical execution is no longer a differentiator.

So the question that remains is: “What should we build?”

Following Paul Graham’s logic, the ability to choose that “what” is exactly what taste is. He made this argument back in 2002, in his essay “Taste for Makers”: “The recipe for great work is very exacting taste, and the ability to execute it.” The reason a 22-year-old essay went viral again in 2024-2025 is that AI has effectively solved the “ability to execute” part.

Venture capitalist Marc Andreessen took this a step further, arguing that once the AI age fully arrives, VC (venture capital) — the skill of picking out valuable investments — will be one of the last professions left standing.Sure, he’s a VC, so there’s an obvious bias at play, but the logic itself is coherent.

The problem is that this logic narrows the word “taste” toward a very specific direction. It starts calling “the ability to pick what makes money” by the name “taste.”

The Two Faces of Taste — Voltaire vs. Silicon Valley

Looking briefly at the history of the concept of taste reveals just how narrow Silicon Valley’s definition really is.

18th-century French philosophers, Voltaire in particular, tied taste to inner human sensibility. “To have taste is not enough to see and know what is beautiful in a given work; one must feel beauty, and be moved by it.” Here, taste is an emotional response as much as an aesthetic judgment.

Pierre Bourdieu² flipped this concept on its head in his 1979 book Distinction (La Distinction). According to him, taste isn’t personal sensibility — it’s a marker of social position. What you find beautiful is determined by your level of education, class, and cultural capital³. Loving classical music counts as “higher” taste only because the opportunity to learn how to appreciate it is concentrated in a particular class.

Set against these two views, Silicon Valley’s “taste” discourse ends up producing a third definition entirely:

Paul Graham’s original essay actually hints at this. He described taste as “being honest with yourself,” yet the context of “great work” throughout the essay was strictly product and business. Poggio co-founder Matt Slotnick made a sharp observation on X: “Taste works as a word because it’s vague, unfalsifiable, and it flatters the ego."

"Taste-Washing” — AI Companies’ Humanist Cosplay

Let’s dig a bit deeper into what Kyle Chayka calls “taste-washing.”

Last year, Anthropic opened a pop-up café in Manhattan and handed out baseball caps embroidered with the word “thinking.” OpenAI’s Super Bowl ad, “You Can Just Build Things,” showed hands gripping bicycle handlebars, hands writing in a notebook, hands playing chess — all shot from a human point of view, even though the product being advertised is, in theory, an omnipotent AI robot.

This is déjà vu from 2010s hipster culture — the era of craft IPAs, American Apparel, and Arcade Fire as identity markers, a fetish for the “lo-fi” and the “handmade.” And in the end, that culture got absorbed by Meta’s Instagram and Amazon’s Whole Foods. As Chayka analyzed in his earlier book Filterworld, algorithms went on to homogenize even that “taste.”

AI companies today are repeating the exact same pattern. They’re selling automation while borrowing the aura of craftsmanship. The message “AI builds it, so all you need is your ‘taste’” is, in effect, a lifestyle marketing pitch: “use our tools, and you too can be a cool person.”

Meanwhile, plenty of people actually experience AI tools as a threat — a threat to their livelihood, a source of anxiety about the future, a challenge to their own identity. The attempt to paper over that gap with the elegant word “taste” — that’s taste-washing.

What the NYT’s Blind Test Reveals — An Uncomfortable Truth

In this context, a recent New York Times experiment is worth a look. On March 9, the NYT placed AI-written passages side by side with human-written ones and asked readers which they preferred. 86,000 people took part, and 54% chose the AI-written piece.

Reading this as “AI now writes better than humans” would be too simplistic. Look at the experimental design: the AI was first fed existing strong human writing before generating its own versions. Writers with distinctive styles, like Cormac McCarthy, were included, and some commenters admitted “I picked the AI because I don’t like McCarthy.” Literary critic Ben Mauk pointed out that “the very act of framing a preference over a short excerpt as a judgment on the ‘work’ itself is the problem.”

But what caught my attention was something else. One blogger’s analogy stuck with me — “the fact that people prefer potato chips over roasted Brussels sprouts in a blind taste test doesn’t mean we should get rid of farmers, chefs, and grocery stores.” Preference and value operate on different planes entirely.

When this experiment collides with Silicon Valley’s “taste” discourse, an irony emerges. If the “taste” they’re talking about is truly aesthetic judgment, how do we explain the fact that more than half of 86,000 NYT readers chose the AI’s writing? Isn’t it possible that the digital ecosystem has already begun reshaping our very capacity to discern taste?

Oz’s Lens

Honestly, there’s one word missing from Silicon Valley’s “taste” discourse that keeps nagging at me: “conviction.”

Ad industry writer Artem Voronov did an interesting analysis on his blog. He tracked the language used by the Cannes Lions 2024-2025 jury presidents when describing award-winning work, and the word “taste” never once appeared. Instead, phrases like “incredibly brave” and “took a massive risk against the critics” showed up repeatedly.

Let’s also revisit Apple’s famous 1984 Super Bowl ad. The board of directors unanimously hated it. They wanted it pulled. But Jay Chiat of the ad agency Chiat/Day deliberately didn’t sell off the airtime, and Wozniak offered to cover the cost out of his own pocket. The result? $3.5 million in sales right after the Macintosh launched, millions of dollars’ worth of free publicity, and a spot as Advertising Age’s “Ad of the Decade.”

This is not a story about taste. It’s a story about conviction — the ability to push forward your own judgment when everyone else says you’re wrong. In my own experience, “the ability to choose what to build” absolutely matters. But it isn’t taste — it’s a combination of a hypothesis about the market and the conviction to back that hypothesis. Taste is vague and unfalsifiable. A hypothesis is concrete and testable. Silicon Valley may have chosen the word “taste” precisely because it wanted to elevate verifiable judgment into an unverifiable realm.

Here’s one more uncomfortable question. There’s already research showing that AI is optimized for flattery. According to a paper published by Anthropic researchers, AI models tend to abandon correct answers under social pressure and align with the user’s opinion instead. This is called “sycophancy,” and the phenomenon worsens as models get larger. Can we really call the output of a tool optimized for flattery “the crystallization of taste”?

Closing

First, Silicon Valley’s “taste” discourse is a new tool for class stratification born from AI’s democratization of production. The narrative that “people with good taste” are the winners of the AI age is, in the end, access to capital and platforms repackaged in aesthetic language.

Second, “taste” is an unfalsifiable word. What Silicon Valley actually believes matters is judgment and conviction — the ability to read data, form a hypothesis, and take on risk. The moment you call that “taste,” a concrete skill gets transformed into a mystical gift.

Third, as the NYT experiment shows, the digital environment is already reshaping our aesthetic judgment. Before declaring “the age of taste,” shouldn’t we first check whether the conditions for taste to actually function still exist?

As Bourdieu showed in 1979, taste has always been the language of power. The same holds true in 2026 — the only difference is that this time, the power flows not from cultural capital, but from computing capital. In other words, what we need is the courage to decide on something and push it through.

References & Further Reading

• Kyle Chayka, “Why Tech Bros Are Now Obsessed with Taste”, The New Yorker, 2026.
• Paul Graham, “Taste for Makers”, 2002.
• Kevin Roose & Stuart A. Thompson, “Who’s a Better Writer: A.I. or Humans?”, The New York Times, March 9, 2026.
• Pierre Bourdieu, Distinction: A Social Critique of the Judgement of Taste, 1979. — A sociological classic showing that taste is not a marker of personal sensibility but of social class. Published in Korean translation as Guboonjitgi.
• Kyle Chayka, Filterworld: How Algorithms Flattened Culture, 2024. — A book on how algorithms homogenize cultural taste, and the backdrop for today’s discussion of “taste-washing.” Published in Korean translation as Pilteowoldeu.
• Anthropic, “Towards Understanding Sycophancy in Language Models”, 2023. — Research on the tendency of AI models to align with user opinions, a phenomenon known as “sycophancy.”

The author, Kwangseob Ahn, is a professor of business administration at Sejong University and lead consultant at OBF (Oswarld Boutique Consulting Firm). He teaches statistics and data analysis — business data management and business analytics — while leading GTM and AI strategy consulting in the field, designing the seam between technology and business. He has published academic research on a memory architecture for AI dialogue systems (HEMA) and runs Daily Arxiv, a daily curation of global AI papers. He holds a master’s from Korea University’s Graduate School of Technology Management and a KMBA. He is the author of Homo Brainless: The People Who Outsource Their Thinking.

Footnotes

1. LLM (Large Language Model): the core technology behind AI systems like ChatGPT and Claude. These models are trained on massive amounts of text data to generate natural-sounding sentences. ↩

2. Pierre Bourdieu (1930-2002): a French sociologist who empirically demonstrated that cultural taste is not a matter of individual choice but a reflection of class structure. The International Sociological Association named his Distinction one of the most important sociology books of the 20th century. ↩

3. Cultural Capital: a concept coined by Bourdieu. The idea that non-economic assets — education, knowledge, taste, manner of speech — determine social status, rather than money (economic capital). Put simply: “the household you grew up in determines the kind of music you come to love.” ↩