AI Reveals Raphael May Not Have Painted All of This Iconic Artwork

AI may have cracked the code to a long-standing question about a famous Raphael painting that has puzzled the art world for years.

Part of Raphael’s “Madonna della Rosa” painting may not have been painted by the Italian Renaissance master, according to findings published in Science Heritage Journal in December 2023, which are now resurfacing and circulating online.

Raffaello Sanzio da Urbino is widely considered one of the most renowned artists of the Renaissance period for his paintings and frescoes, including the “School of Athens” in the Vatican.

But the “Madonna della Rosa,” which is on display at the Museo Nacional del Prado in Madrid, has long sparked speculation about its attribution, with the image of St. John appearing inconsistent with Raphael’s style and less refined compared to other elements of the painting. The painting shows Madonna and Baby Jesus alongside St. Joseph and St. John. 

Computer scientists at the Universities of Nottingham, Bradford and Stanford developed a specific AI algorithm to compare parts of the painting to other Raphael pieces.

The team used pictures of verified Raphael paintings to “train the computer to recognize his style to a very detailed degree, from the brushstrokes, the color palette and the shading and every aspect of the work,” according to Hassan Ugail, a computer scientist from the University of Bradford.

“When we tested the della Rosa as a whole, the results were not conclusive,” Ugail said in a statement. “So, then we tested the individual parts and while the rest of the picture was confirmed as Raphael, Joseph’s face came up as most likely not Raphael.”

In a press release, the computer scientists said it is believed one of Raphael’s pupils, Giulio Romano, likely painted St. John’s face. Although the findings aren’t 100% confirmed, Ugail noted the computer model examines artwork on a microscopic level with a 98% accuracy rate. 

Christopher Brooke, honorary research fellow at the University of Nottingham who was also involved in the project, said in the press release the analysis “promises to be a useful, additional tool in future investigations of this nature, alongside well-established methods such as spectroscopy.”