Battle of the AI: rival tech groups clash over who painted ‘Raphael’ in a UK gallery

Authenticating works of art is not an exact science, but the painting of the Madonna and child has sparked a fierce controversy, dubbed the “war of AI,” after two Separate scientific studies reach conflicting conclusions.

Both studies used modern AI technology. Months after a study claimed that the so-called by Brecy Tondocurrently on display at Bradford council’s Cartwright Hall Art Gallery, is “definitely” by Raphael, another has found that it cannot be by the Renaissance master.

In January, research teams from the universities of Nottingham and Bradford published findings on facial recognition technology, which compares faces in Tondo with the people in Raphael’s Sistine Madonna altar, put into use in 1512.

After using “millions of faces to train an algorithm to recognize and compare facial features,” they claimed: “The similarity between the madonnas found was 97%, while comparing The child in both pictures shows 86% similarity.

“This means that the two paintings were most likely created by the same artist,” they added.

But algorithms in a new study by Dr. Carina Popovici, a scientist with Art Certification, a Swiss company based near Zurich, have now returned an 85% probability that the painting was not painted by Raphael. .

The Tondo purchased by Cheshire in 1981 businessman George Lester Winward, who built an art collection spanning the 16th to the 19th centuries. In 1995, two years before his death, he founded the de Brécy Trust Collection, named after the name of his French ancestors, to preserve his collection and make it available to scholars for research.

The painting has been subjected to extensive examination and historical research over more than 40 years.

In July, Professor Hassan Ugail, director of the Center for Visual Computing at the University of Bradford – who developed the AI ​​facial recognition system – said: “My AI models look deeply into a photo than the human eye, comparing details such as strokes and pigmentation.

“Along with my previous work using facial recognition and combined with the previous research of my fellow scholars, we concluded Tondo and Sistine Madonna definitely by the same artist.”

Professor Christopher Brooke of the University of Nottingham and a historian of ecclesiastical art, said: “This research demonstrates the ability of machine learning to provide probabilities of the same artist between ‘Old’ paintings Masters’ are different. In this case study, direct facial comparison was 97% accurate – a very high statistical probability that the artworks are by identical creators.”

The analysis of pigments is also said to place it firmly in the Renaissance.

However, Popovici was surprised to discover that the results of her study “clearly contradicted” their analysis in a significant way.

The art recognition has an ongoing collaboration with Tilburg University in the Netherlands and its research was recently published by Springer, the academic publisher. It analyzed more than 500 works, including a painting by Rubens in the National Gallery – View of Het Steen in the early morning – is given with a probability of 98.76% in favor of the artist.

A spokesman for Bradford council, where the painting hangs in the building, joked: “I guess it’s a war of AIs.”

Sir Timothy Clifford, a leading scholar of the Italian Renaissance and former director general of the National Galleries of Scotland, was intrigued to hear about the conflicting findings in science: “I actually feel quite strongly about It is clear that mechanical means of identifying paintings by great artists are extremely dangerous.

“I never thought about the idea of ​​using these AI things. I think they are very unlikely to be remotely accurate. But how fascinating.”

When shown a small photo of Tondo, he believes that it is an extremely authentic copy of one of the most famous paintings in the world: “There must be hundreds and hundreds of very good copies of it. My immediate reaction was that it was probably a French, early 19th century, good copy – perhaps around the time that the painting I’m sure went with Napoleon’s spoils of war to the Louvre .

“Many of the pigments used in 1810 were the same pigments that were used in 1510. Unless one or two of those were being used, which should not have been used, The results will be very similar. .”

Michael Daley, director of ArtWatch UK, which monitors museums and galleries, said: “It’s good that Carina Popovici has opposed it. Its only claim, as far as I understand it, is to have a perfect correspondence with the design of Sistine Madonna.

“However, if Raphael himself had created another version, he would almost certainly have introduced his own modifications or deviations to the painting. Almost certainly 19th century.”

When informed of the conflicting research, Ugail said that without seeing the details it was difficult to comment: “We have a very strong case to prove that [the Tondo] is a Raphael.”