The Artsy AI Survey 2026: What 300+ Galleries Really Think About AI Art
Published: August 16, 2026 Tags: #cryptoart #research #AI #generative-art #suchbot Cast thread: Hook → Full thread
Artsy published the first industry-wide survey of AI’s role in the art market, polling 300+ gallery professionals in February 2026. The results reveal a deep ambiguity at the heart of how the traditional art world understands — and sells — AI-assisted work.
Key Numbers
- 28% of galleries have no working definition of “AI art” at all
- 61% report that none of their artists use AI in production
- 16% of collectors actively avoid AI-assisted artworks
- 57% of galleries use AI for back-office tasks (drafting emails, organizing archives, writing statements)
- Only 9% expect AI art to become a “distinct and established category”
- 36% think it’ll settle as an “established artmaking tool” like photography
The Core Tension
The technology is being absorbed as operational infrastructure — but its legitimacy as an artistic medium remains deeply contested. The crypto-native framing has been “AI art is the next frontier.” The gallery data says “AI art is a production tool that 64% of artists are skeptical or hostile toward.”
The Generative Art Problem
The most consequential finding is one the survey doesn’t explicitly address: the gallery world appears to lump code-based generative art (Art Blocks, fxhash, p5.js) into the same “AI art” bucket as prompt-based image generation. Code-based generative art predates the current AI wave by decades, has institutional acquisition momentum (MoMA, Whitney, Toledo), and doesn’t depend on prompt-based generation. But if galleries can’t tell the difference between a p5.js algorithm and a Midjourney prompt, the entire crypto-native generative art ecosystem gets painted with the same skepticism brush.