Why tech bros are so worried about AI having bad taste

Published
Score
9

Why it matters

Tech leaders, particularly "tech bros," are increasingly emphasizing "taste"—human judgment in curating and directing AI outputs—as a vital skill amid fears that AI generates generic, low-quality "slop" lacking aesthetic subtlety or uniqueness.[1][2][4] This discourse escalated with public statements from figures like Y Combinator cofounder Paul Graham (who posted that "taste will become even more important" in the AI age), OpenAI president Greg Brockman (declaring "Taste is a new core skill"), Framer founder Koen Bok, and Cloudflare CTO Dane Knecht (calling taste the "engineering differentiator" for 2026).[4] Critics like Linear's Nan Yu, essayist Matt Shumer, and Wharton professor Ethan Mollick counter that AI can learn taste via training on signals like citations, upvotes, or personal data, as shown in a March 2026 paper where models predicted influential research papers.[2]

The trend stems from earlier reporting, such as New Yorker writer Kyle Chayka's 2018 essay "Style Is an Algorithm" and his recent coinage of "taste-washing"—ascribing human-like qualities to AI to mask its flaws—and a prior piece on tech's taste obsession.[4] It gained traction in early 2026 via X posts and memes mocking Silicon Valley style (e.g., Allbirds, quarter-zips), alongside actions like Anthropic's "Zero Slop Zone" pop-up and Mark Zuckerberg attending Prada.[4] Experiments confirm AI's limits: even GPT-4 achieved only 50-65% accuracy predicting personal preferences from user data.[2]

Newsworthy now due to accelerating AI automation threats—like "AI-washing" layoffs without infrastructure (per Forrester)—and ironic debates where tech elites prize taste while AI advances in idea generation and curation.[4] As of April 2026, this highlights tensions between human uniqueness and machine replication in creative fields.[2][4]

Sources

mail

Get notified about new Artificial Intelligence developments

Primary sources. No fluff. Straight to your inbox.

See more entries tagged Artificial Intelligence.