AI and the Art of Cultural Memory

On 23 March 2023, Sotheby’s New York auctioned Edmond de Belamy, an AI-generated portrait by the Paris-based art collective Obvious, for $432,500—ten times its high estimate. This sale underscored AI's cultural significance and sparked debates about authenticity. Who is the artist? Does it matter?

AI-generated art occupies a complex space. When German artist Mario Klingemann won the Lumen Prize in 2018 for Memories of Passersby I, he described it as a "collaboration with the machine." This contrasts with platforms like DALL-E or MidJourney, where millions create art through prompts. The distinction lies in intent and labor—an area courts have yet to fully define. In August 2023, the US Copyright Office reaffirmed that works lacking substantial human authorship are public domain.

Cultural preservation raises further stakes. In 2020, Google Arts & Culture digitally reconstructed The Ghent Altarpiece by Jan van Eyck, whose lower panels were irreparably damaged. Critics like art historian Dr. Bendor Grosvenor question the authenticity of such restorations, warning that AI could 'overwrite' human history with approximations.

AI also aids architectural conservation. Neural radiance fields (NeRF) technology enabled 3D reconstructions of Syrian heritage sites destroyed during the civil war. In 2021, the Centre for Cultural Heritage Technology in Venice partnered with UNESCO to model damaged frescoes in Aleppo’s Khuwayja Mosque. Dr. Gabriella Tesorieri, the project’s lead researcher, told ARTDESENT, "The reconstructions bridge archival gaps, but are no substitute for the originals."

Artists critique AI's colonial implications. American-Mexican artist Karen Navarro’s Latent States generates speculative portraits of pre-Columbian figures erased from colonial records. In an interview with Hyperallergic, Navarro stated, "Datasets mirror power structures; training them is, itself, an act of politics." This aligns with digital anthropologist Dr. Ruha Benjamin’s assertion that "training algorithms could perpetuate cultural erasure under the guise of innovation" (Race After Technology, 2019).

Ethical considerations extend to data sourcing. Generative AI often scrapes copyrighted images without consent. In February 2023, three visual artists filed a class-action lawsuit against Stability AI, claiming unlawful use of their images for training. Such cases could set precedents for intellectual property in machine learning. The European Union’s AI Act, set for enforcement in 2026, will address data transparency to mitigate conflicts.

Some view AI as a chance to democratize art. Initiatives like MoMAR, which “hacks” museum exhibitions with augmented reality, challenge institutional gatekeeping. London’s Design Museum hosted AI: More Than Human in 2019, exploring these disruptions. Curator Anna Ridler noted that “industrialisation faced similar resistance,” drawing parallels between today’s AI anxieties and 19th-century fears over mechanized looms.

Art’s relationship with AI remains contested. Walter Benjamin’s idea of aura—the uniqueness of art tied to its history—faces scrutiny. When AI generates a 'lost' Vermeer, does its lack of provenance render it soulless? Or does the act of reconstruction itself hold value? These questions remain unresolved.

AI emphasizes art as process, not product. Its ability to iterate shifts focus from finished pieces to the infrastructures shaping culture. Museums must grapple with digital preservation strategies, not only for physical works but also for codebases. Artists must interrogate the systems they create or challenge. As Tesorieri observes, "Whether AI erases or enriches cultural memory depends on whose hands it’s in."