The art market has always known its own secret: you are not buying the object, you are buying the provenance. This week, two stories arrived to finally put numbers on it.

When AI Exposes the Market's Core Fiction

ARTnews reports that AI valuation models consistently rank unknown abstract works above canonically famous ones when stripped of attribution data. The model isn't wrong, it's just measuring the wrong thing. The art market doesn't price paint. It prices biography, estate management, gallery relationships, and auction house theater. A 2025 study would frame this as pure Veblen dynamics, but the AI result is actually more devastating: it suggests the market's pricing logic is entirely detached from any aesthetic signal the object itself emits. Simultaneously, Yale's Sotheby's sale raising over $1M for a debt-free MFA program shows the flip side. When artists with institutional pedigree donate works, collectors pay, not for the canvas, but for the credential chain: Yale, Sotheby's, and the social capital of participating in institutional virtue. , and the art valuation story is a perfect case study for where AI surfaces uncomfortable truths that human systems have a structural interest in suppressing.

Old Masters, Unknown Hands, and Market Mythology

Artnet's Old Masters roundup notes that one of 2025's top-selling works is by an unknown artist. Which means the market is now rewarding mystery as a category, the unknown as a brand. Read alongside the AI valuation paradox, a pattern crystallizes: the market rewards legible narrative, whether that narrative is a famous name, an institution, or a compelling absence. The AI doesn't understand narrative. That's why it keeps picking the wrong Picasso. This question of who defines cultural value, and what infrastructures sustain those definitions, sits at the heart of what Max Hollein has articulated about museums navigating the digital age: taste is a system, not a property of objects.