The two-month exhibition at the Cartwright Hall Art Gallery will show how AI models will explore the field of professional art attribution, traditionally the domain of specialists who can devote their entire careers to the study of a single artist or era. It shows how a small world can be turned upside down.
Researchers from the University of Bradford and the University of Nottingham said in January An AI-powered facial recognition model shows a 97 percent resemblance between the Virgin Mary depicted in the controversial painting by de Blessy Tondo and the Sistine Madonna identified by Raphael; They also found an 86 percent resemblance between the child and the altarpiece’s Jesus. Some experts on the artist disagree.
Hassan Ugair, a professor of visual computing at the University of Bradford who developed the model, said the case showed it was time for art lovers to modernize their toolkits.
“It was quite a learning curve to understand the art world and how it uses very little scientific evidence,” Ugair said. Researchers in the model zeroed in on “dimensions that are invisible to the human eye,” he said.
Richard Polsky, who runs an authentication company focused on 20th-century American artists, said art attribution can come with huge financial stakes.
“Suppose there are very few authentic pieces in museum collections. They have been in private hands for centuries, so adding one more to the market could fetch a huge price. ” Polsky said. “It doesn’t really take someone with money to say, ‘I want this in a new museum and I’ll pay $100 million.'”
Some artists have a definitive catalog of attributed works. Some people don’t. This means that academics and the art market need to reach consensus on new works that have the potential to sell for millions of dollars and change audiences’ understanding of artists.
Proponents of using AI in this process argue that it makes the attribution process fairer.
“Relying on the judgment of a single human expert can be risky because of the potential for human error, subjectivity, and bias,” said De Blessy Tondo, who was not involved in the case. said Karina Popovic, CEO of AI recognition company Art Recognition. study. She said her group’s algorithm accurately sniffs out paintings by forger Wolfgang Beltracchi. Admitted to forging works in 2012 Works by approximately 50 artists.
However, in the case of de Blessy Tondo, Raphael and other Renaissance art experts say they are skeptical of the AI model’s results.
personal collector I bought the painting in 1981., noticed its similarities to the Sistine Madonna, and was transferred to a trust in 1995 for experts to study. Supporters call it Raphael’s original. Others say this is a copy made during the artist’s lifetime. Nineteenth century.
Rudolf Hiller von Gardlingen, an art historian at the University of Leipzig and co-author of a comprehensive collection of Raphael’s works, said that at the height of Raphael’s career, Raphael made a copy of the figure depicted in the Sistine Madonna. He said he didn’t think he did. In the 1510s he preferred to change subjects.
Patricia Emison, an expert on the Italian Renaissance at the University of New Hampshire, agrees.
“Redrawing the Virgin and Child motif seen in movies” [Sistine Madonna] “The altarpiece is below his artistic dignity,” Emison said. “He’s not just looking for extra money.”
Polsky and Emison said they are unsure whether AI models can detect the subtle nuances that experts see when looking at works of art.
“If you’re deeply immersed in an artist’s work, you’ve read everything about that artist. You go to museums around the world to see original works, you go to gallery exhibitions, you go to some You may have owned one or bought or sold one,” Polsky said. “I don’t think you can teach something like that.”
Despite skepticism from traditional arts experts, AI is rapidly gaining ground in the art world.
AI-driven models are not only differentiating between fake and authentic artwork, they are also creating new art and controversy.A group of artists sued multiple AI companies this yearclaimed that using online art as training material violated copyright law.
Unlike litigation, the final decision is De Blessy Tondo’s origins can be difficult to know.
“Attribution is nuanced. It’s not something you can practice with absolute certainty,” Emison said.