The news that Lee Krasner will make her Paris debut via Gagosian and Olney Gleason this October, timed to a high-profile Met exhibition, arrived the same week that Hyperallergic was asking who, exactly, asked for an AI art museum. Both questions are secretly the same question: who decides which work gets preserved, surfaced, and institutionally validated, and by what mechanism?
Institutional Memory as a Curatorial Problem
Krasner died in 1984. Her Paris debut is happening in 2026. That gap is not explained by quality. It is explained by the structure of the art world's attention economy, the same economy that a new AI art museum implicitly promises to disrupt or democratize. But the Hyperallergic skepticism is well-founded: AI curation trained on existing institutional data will reproduce existing institutional biases at scale and speed. The algorithm of erasure does not disappear when you automate it. It accelerates. A Naples show pairing Adam Pendleton with Antoni Tàpies is doing something more interesting by hand: forcing a confrontation between two eras of abstraction that the market would never naturally produce.
Archives, Attention, and What Gets Surfaced
The deeper throughline here is archival. David Senior of SFMOMA's Library has written about how the choice of what to preserve is itself a curatorial act with aesthetic and political consequences. Krasner's belated Paris debut is an archival correction, the kind of restoration that requires human institutional will, not algorithmic recommendation. The AI art museum question is really asking whether we trust systems trained on the canon to revise the canon. The answer implicit in Krasner's story is no.