Refik Anadol to Project AI-Generated Work Onto the Smithsonian Castle
Fresh from the opening of Dataland in Los Angeles, Turkish media artist Refik Anadol (b. 1985) is turning to one of the most recognizable buildings in American museum culture. On July 17 and July 18 at 9 p.m. he will present “Smithsonian Dreams,” an immersive installation projected onto the Smithsonian Castle on the National Mall in Washington, D.C.
The project is being created in partnership with the Smithsonian Institution and draws on a custom AI system trained on 200 years of Smithsonian collections and research. That material includes millions of digitized specimens, manuscripts, photographs, artworks, objects, and scientific records. Anadol’s system reprocesses the archive into light, sound, and visuals, which will be mapped across the Castle’s facade through a custom visualization system using a UMAP algorithm.
The structure of the work will sound familiar to anyone who followed “Unsupervised,” Anadol’s much discussed installation at the Museum of Modern Art. That project also used a museum collection as raw material for a machine-generated visual environment, with curatorial language describing the output as a form of dreaming. According to the artist’s studio, about three million people visited MoMA to see it, spending an average of 38 minutes in front of the piece.
In a statement announcing the Smithsonian project, Anadol described data as “a form of memory” and said the work asks what might emerge if the institution’s vast archive could become dynamic, reflective, and alive to its own history. He framed machine intelligence as a creative collaborator rather than a replacement for human interpretation.
Whether viewers see the result as a poetic reimagining of institutional memory or as a highly polished digital spectacle, the Smithsonian commission underscores how central AI has become to contemporary art’s public-facing ambitions. It also places one of the country’s most storied repositories of knowledge inside a visual language built for scale, immersion, and mass attention.
