Legendary Italian master Raphael's "Madonna della Rosa," otherwise known as "Madonna of the Rose," has long vexed both art connoisseurs and historians for hundreds of years. However, due to a recent discovery made by an AI algorithm, that mystery is finally laid to rest.
This breakthrough was featured in a new paper published in the open-access journal "Heritage Science," last Thursday, Dec. 21, by researchers from the University of Bradford and Stanford University.
'Madonna della Rosa' and Its History With Skepticism
The painting itself depicts a biblical scene involving Mary, Joseph, and infant depictions of both Jesus and John the Baptist. It was solely attributed to Raphael up until the 19th century when art historians began to argue that artists from his workshop should also be credited for the work.
Initially, it was because the depiction of Joseph's face looked as if it was painted as an "afterthought" and exhibited characteristics that were not consistent with how Raphael typically painted his works. Other scholars were also skeptical that it was a solo painting, but they fixated more on the rose that was shown on the lower part of the piece.
Despite all the conjecture, the painting was attributed to only Raphael for the longest time, as it was hung in Museo del Prado's halls. However, when the piece was examined by an AI developed by Professor Hassan Ugail, Director of the Centre for Visual Computing at Bradford, it turned out former art scholars were right on the mark.
Specifically, regarding the face of Joseph being painted by a different artist, consequently, the lower portion was confirmed to still be made by Raphael's hand.
How the AI Confirmed the Painting's Authenticity as a Raphael Piece
Ugail told The Guardian that the algorithm was trained using 49 works that were undoubtedly painted by Raphael, and as a result, is now able to determine whether works are genuinely made by the Italian master with a 98% certainty.
The Bradford professor added that the AI meticulously looks at every part of the painting in a "microscopic way," saying, "Not just the face, it is looking at all its parts and is learning about the color palette, the hues, the tonal values, and the brushstrokes."
In the case of Raphael's "Madonna," the first few tests have revealed that 60% of the painting was authentically made by the Italian painter. The AI, through its sectioned investigation, eventually determined that the part with Joseph's face wasn't made by Raphael.
Howell Edwards, emeritus member of faculty for the Molecular Spectroscopy department in Bradford and co-author of the study, said that the AI program has picked up where their previous work has been stumped and has thus "unequivocally" determined what was made by Raphael, and what was not.
AI as a Tool for Authenticating Art
Another widely debated painting related to Raphael was also tested using the AI program. The piece was named "Haddo Madonna" due to where it is currently located, which is in Aberdeenshire's Haddo House.
The piece was originally purchased in the 19th century under the pretense that it was a genuine Raphael piece by George Gordon, prime minister of Britain between the years 1852 to 1855. The result of the test has confirmed its authenticity as a Raphael work, a notion the National Gallery experts have denied in 2016, according to Daily Mail.
Ugail himself admits that his vocation lies elsewhere, away from art, which could arm the possible skepticism that art scholars have with his AI's track record. Nonetheless, the Bradford professor still holds out hope that those scholars will eventually be convinced and see the possibilities of AI as a useful tool to authenticate art.
"The process of authenticating a work of involves looking at many aspects ... its provenance, pigments, condition of the work, and so on. This sort of software can be used as one tool to assist in the process," he said.
