Taste has long served as a mechanism for legitimizing power and authority.

This concept of “taste as a moral technology” traces back as far as 18th-century Britain, where the Third Earl of Shaftesbury argued that to cultivate taste is to cultivate personal virtue. Every few decades, some debate over what constitutes 'tasteful' aesthetics resurfaces, spanning from the battle between geometric modernism and historical ornamentation, to the material candor of the Good Design movement, to Bourdieu’s exposure of taste as cultural capital. Today, Silicon Valley tech bros have embraced the taste discourse as their latest plat du jours (pun intended). In a widely discussed New Yorker essay examining the trend, Kyle Chayka notes that “this emphasis on tasteful decision making makes sense, given that AI is gradually democratizing technological production.”

Many industry leaders have echoed this sentiment: Marc Andreessen told Fortune that the curatorial art of picking (read: investing) may be among the few roles safe from automation. ByteDance's Cong Wang similarly contends that personal taste forms a defensible moat. Venture capitalists and founders alike have eagerly pointed to Paul Graham’s assertion that “in the AI age, taste will become even more important,” or Greg Brockman’s tweet calling taste “the new core skill,” (a phrase popularized by artist Claire Silver as early as 2024).

Imagine a brand’s identity structured as a three-tier pyramid. The top tier consists of surface-level elements like fonts, colors, and logos that visually identify the brand. The middle tier contains the design system, a framework defining how to deploy those graphic elements. At the foundation lies a rare, intangible element: something like experience or intuition, which serves as the root of the entire visual ecosystem. This foundational layer might be built on years of accumulated judgment, shared history, cultural references, and lived knowledge that directly shape how audiences perceive and interact with a brand.

In an attempt to align themselves with taste, many tech companies fall into the formidable trap of extracting top-layer visual assets to build middle-layer design systems, ultimately simulating, or “larping,” the bottom layer. By now we’ve seen countless hand-waving efforts to automate taste from startups promoting tools for “creative orchestration.” The premise is simple: input any URL, and their software can instantly mirror the brand's aesthetic, from its landing page copy to its font library, UI components, and color palettes, organizing everything into a neat directory. This is “Design at the speed of AI.”

But these mechanisms only automate the output of taste, focusing entirely on the signifiers at the apex of the pyramid—the parts that can be easily extracted, repackaged, and sold back as genuine creative production. By focusing solely on these surface artifacts, we abandon the foundational elements that actually engender and cultivate taste: the interplay between doubt and desire, the visceral intuition, gut checks, dialogue, calculated risks, disruptions, and labored research required to translate inherited culture into resonant expression. None of those vital qualities can be captured in a brand kit.

With the proliferation of AI imagery, the cost of producing surface level visual assets is hurtling toward zero. The shift mirrorsToby Shorin’s 2018 diagnosis of the diminishing marginal value of aesthetics: as the cost of producing and distributing a particular aesthetic falls, so too does its ability to register as novel. Today, virtually every aesthetic dialect is instantly accessible to everyone.

Consequently, (for whatever reason) tech culture’s understanding of taste has dwindled into a narrow canon of repeatable tropes: the minimalist functionalism of Dieter Rams, glossy Taschen art books, dystopian cyberpunk aesthetics, neoclassical iconography, lingering remnants of Corporate Memphis. Tech leaders tend to lean on these motifs to project an aura of cultural awareness, capital, influence, and intellectual gravitas, each actor affirming with fervent head nods that taste matters.

Writing in The Creative Technologist from Baukunst , Shorin observes a curious dynamic revolving around contemporary tech circles:

“There’s a weird cultural equilibrium right now where you can get hundreds of likes by tweeting something generic like: ‘how come nobody dresses cool and tasteful in SF? We must change this’ with a mediocre fit pic.”

Emily Segal of Nemesis characterized this homogenizing impulse as tasteslop: the inevitable outcome when superficial indicators of taste are stripped from their original contexts and recycled, amounting to “cultural capital after extraction, after it’s been through the blender.”

The act of co-opting cultural capital is certainly not new. However, generative models have drastically lowered the barrier to entry. AI can now harvest, synthesize, and redistribute emerging cultural markers as quickly as they form. The temporal lag between aesthetic novelty and widespread saturation, a window that once spanned years, now collapses over the course of days, or even hours.

A handful of startups have made a business of formalizing what taste actually is, each promising some version of the same thing: that aesthetic judgment can be broken down, extracted, priced, and delivered as a service for brand building.

• Taste Labs’ latest “Brand API” markets itself around three core functions: extract a brand system from what a client already has, search a curated index for inspiration, and verify new work against the extracted rules, treating brand identity as a complete corpus, ready to dissect

• Surge AI, which supplies post-training data to OpenAI, Anthropic, Meta, and Google, recruits poets alongside mathematicians, Supreme Court litigators, and former startup CEOs, treating taste as a credentialed domain to be encoded and tracked. It markets its mission to encode AGI with the richness of humanity: curious, witty, imaginative, and wise. Its own product page describes capturing “the notes that separate the ordinary from the inspired, Salieri from Mozart.”

• Contra Labs is a data evaluation lab that operationalizes brand by leveraging a vetted network of independent creative professionals to evaluate AI outputs across multi-stage workflows and scoring models against specific criteria (like design system adherence). Their own co-founder, however, has said plainly that he does not believe it is possible to give a model taste at all: “LLMs can improve their outputs based on that preference data, but they're never going to be able to create something truly tasteful and competitive that's net new and resonant...”

Each of these models relies on a fundamental misunderstanding of what a “brand” actually comprises. A brand can exist long before any website launch, and it encompasses far more than a set of fonts, colors, or visual guidelines. Returning to the pyramid model, the most interesting brands originate from some intangible core foundation—a wellspring of inspiration that informs all upstream creative output.

Speaking to his experience as an ECD at Complex, Air's Brand Director, Wyeth Hansen shares:

“Brands need to be elastic, and porous. A good-looking but rigid brand doesn't equal affinity—first, you have to make things that people genuinely love. Previously, I was at Complex, and we had rabid audiences for Hot Ones and Sneaker Shopping, among others. There was a lot of love for those properties even if they weren't necessarily following core brand guidelines. What that showed me is that it doesn't matter if the colors and typeface are on-brand if no one is paying attention. It's entirely possible to brand yourself into irrelevance. At

Air, we're guided by an ethos and simple visual touchpoints, and the rest is the collective imagination of the team.”

There is no elusive “brand essence” waiting to be discovered, or traced. If you lack a core brand framework—one borne out of opinions, perspective, and experience—then generating ad-hoc collateral via automated tools isn't executing a brand strategy at all. It’s larping on vibes.

Paul Jun, Head of Brand at Ramp, explores this idea in his recent essay, “Brand as Software.” He suggests that while routine production workflows can be automated, identity-defining work built on judgment and conviction remains uniquely human. For Jun, a resilient brand is like jazz: it maintains a clear throughline while leaving room to adapt and improvise. As he puts it: “Taste remains human. Repetition becomes software.”

The latest “Brand as API” software trend mistakes the brand for a triadic workflow, or finished artifact, when it is in fact an ongoing, lived negotiation.

Shorin maps cultural aesthetics across a distinct continuum: the far right signifies edgy aesthetics defined by their obscurity and novelty, while the far left represents universal legibility. This captures styles so thoroughly absorbed into the mainstream that they start to feel obsolete (think: Common Projects, once the IYKYK status symbol of the creative class).


Rather than a linear spectrum, I’ve often pictured this schema as a series of concentric circles. The outer perimeter is where creative momentum originates. This is the domain of underground nightlife; queer culture; subcultural DIY movements; punk nihilism; avant-garde cinema; ballroom communities; radical countercultural traditions. Those on the margins have a distinct perspective: they see the mainstream so clearly precisely because they must navigate its boundaries while building alternative worlds outside of them.

The space between the core and the edge represents the sweet spot for brand building, the delicate terrain where the familiar (read: “normal”) meets the strange and captivating. This is where effective marketing thrives: taking subcultural markers from the margins, aestheticizing them, and converting them into scalable consumer products.

The issue with many of these automated design tools stems from their over-reliance on computational logic. By definition, algorithms lean toward common-denominator outputs, often collapsing into a homogenized average (Hito Steyerl outlines this kind of phenomenon at length in her writing on “Mean Images”). Paradoxically, AI companies try to align themselves with an aura of craft or counterculture, even as their tools threaten to automate human creativity into oblivion.

Embodied creativity precedes structural knowledge; it is almost a proto-knowledge. It relies on raw instinct, intuitive judgment. Where automated systems thrive on frictionless scale and tidy datasets, only humans can engage in the frictionful, bidirectional (dare I say illogical?) thinking inherent to genuine creative work. Speed was never the defining metric for creativity. Rome wasn’t built in a day. The true risk for creatives isn’t getting “left behind,” but becoming generic. 

“How to Develop Taste,” by Ruby Justice Thelot

“Simulating Authenticity in Times of AI,” a conversation between FKA twigs and Mumi Haiati

“Tasteslop is turning Culture into a Moodboard,” by MJ Calayan

“The End of Research,” by Michelle Santiago Cortes

“AI Has Come for Serif Fonts,” by John Semley

Retrograde Prometheus: Subjectivity & Computation by Christian Nirvana Damato

Filterworld, by Kyle Chayka

“Epic Fury: the Dark Art of Defense Tech” by Simon Denny