I have a problem with my Instagram algorithm. I’m a fan of the platform; I’ve been using it for more than a decade to follow designers, architects, and artists. It’s the best mainstream tool for sharing visual culture, and thus unavoidable for anyone in that industry. The feed often shows me new things that I’m interested in: new-to-me modernist painters or furniture brands. But the onslaught of content from the platform has lately turned things that I once thought I liked, or even loved, into boring sameness (I love Mark Rothko paintings; I don’t love a hundred of them in a row in my Discover page). This process of aesthetic satiation—an overabundance or overexposure of something rendering it meaningless—happened to me recently with a certain kind of Mexico City modern interior. You know the type: minimalist, adobe-ish walls; texture by way of fiber wall hangings and handmade ceramics; Axel Vervoordt-style deep couches sprinkled with midcentury pieces. On its face, this combination is quite tasteful. But now, after Instagram has shown it to me over and over again, I hate it.
Good taste becomes bad taste when processed through the engine of the internet. It happens even with the most rarefied of culture. Take Ricardo Bofill’s La Fábrica, for example, the cement factory outside Barcelona that Bofill turned into a family home and studio beginning in 1973. It’s a project of Brutalist adaptive reuse, willfully alienating in its embrace of raw concrete and industrial ruins. It is utterly singular and irreplicable. And yet, I’ve seen it so many times in feeds that I never want to see it again. Once a niche reference point, La Fábrica is now a bit like Kahlil Gibran’s book The Prophet or a Haruki Murakami novel: You cite it as a marker of your taste only if you don’t know that everyone else has already gone through it.
Algorithmic feeds on digital platforms attempt to serve us content that we like; they filter in search of what an equation predicts will be interesting to us. Yet by guessing in advance what we want, they tend to homogenize what we consume. An algorithm will group you in with millions of other users, slotting your tastes into an easy category, then overdeliver on that category. The effect is the same on Instagram, Netflix, Spotify, or TikTok. This ecosystem homogenizes culture by encouraging consumers to consume similar kinds of content as each other, and forces creators to cater to that average bottom line, molding their work or even censoring themselves in order to find an online audience. The majority of culture now flows through one algorithm or another, so whether you make furniture or paintings or buildings or albums, you have to contend with their effects. You have to know what succeeds in the algorithmic space without becoming totally flattened or made cliché by it.
Generative artificial intelligence is the successor to the algorithmic feed. Ever since ChatGPT launched to the public in 2022, chatbots have become increasingly influential filters for information and thus for culture. Yet A.I. is, if anything, more homogenous than the algorithmic social media era. Algorithms pushed creators to publish generic content to appeal to the mass audience, a process that at least had a human element. A.I. produces the generic content automatically, accelerating the cycle of tastelessness.
This is not a hypothesis; it is a fact. Scientific experiments have shown that people who use AI as a writing tool tend to produce writing that is less creative and original than those using only their own brains or simple Google Search, as well as more similar to other people also using AI. The vocabulary of AI writing is homogenized. Extrapolate that to aesthetics, and it means that AI is pushing the same visual reference points and tropes to everyone—an extreme intensification of the Ricardo Bofill problem. This recently played out in the world of web design when users discovered that Claude Design was putting the same orange color, the same serif typeface, and the same rounded button shape in many of the sites that it produced, a pattern that was only clear after hundreds of users posted their work online and noticed the similarity. Individually, these designs might look tasteful, even elegant. They fulfill the requirements of the prompt. But viewed in bulk, they are ugly, repetitive, and boring.
As an exercise, I recently asked ChatGPT to design me a room that a crowd of design executives would find tasteful. What it spat back out was that exact Mexico City-modern mush: blank, textured walls; big wabi-sabi ceramics; Eames Lounge Chair. Too ubiquitous, too late.
A key function of taste is to be a differentiator. Taste discerns between the common and the uncommon, the unique and the homogenized. If basic markers of good taste become banal, then taste must find other targets. In other words, we must consider which qualities escape the trap of digital amplification that I’ve described here. (Otherwise, how can you be sure what you think looks good isn’t exactly what hundreds of other creatives are also being spoonfed by technology?) Is the only good taste then bad taste? Instead of seeking out what is charming and compelling, should tastemakers pursue what is ugly and discordant, because then it won’t be so easily digested?
Lately, the word “texture” has stuck in my mind as a representation of everything that is beyond the algorithm, beyond the machine. Texture means all the little details and incontrovertible facts that are part of unchangeable, unscalable physical reality. The provenance of a piece of art. The history of a building. The lived experience of a space’s designer or its inhabitants. Time, labor, craft, thought. These are the things that AI promises to shortcut or short circuit. Ironically, this automation will only make deep, authentic creativity scarcer and more valuable—if only we can still practice it.
Kyle Chayka is a staff writer at The New Yorker, where he writes a weekly column about technology and culture called Infinite Scroll. His work examines the growing influence of algorithms on aesthetics, attention, and taste, a theme he presents in his 2024 book, Filterworld: How Algorithms Flattened Culture.
This story is from Issue 9 of The Quarterly. Click here to see the full issue.
