- AI company Art Recognition has authenticated over 500 works of art
- As a result, the value of the painting can increase (or decrease) significantly.
- However, this technique has proven unpopular among art historians and curators.
The rise of artificial intelligence has broken new ground in identifying whether works of art are valuable or fake, but the results are met with skepticism among experts. Masu.
Swiss-based AI company Art Recognition has completed more than 500 authenticity assessments, including a high-profile verification of Vincent van Gogh's 1889 self-portrait, the Financial Times reported.
Founded just five years ago, Art Recognition uses an AI system to provide accurate and objective authenticity assessments of works of art.
The company was embroiled in a furor last year over a painting known as “De Blessy Tondo,” which has been at the center of a bitter controversy for more than 30 years.
Art Recognition's AI concluded that there is an 85% chance that the painting is not the real Renaissance master Raphael, as some experts had previously claimed.
This conclusion came despite two British universities, the University of Bradford and the University of Nottingham, using their own AI software to determine authenticity.
The university deployed facial recognition AI software and concluded that the face in the work was identical to the face in another of Raphael's paintings, The Sistine Madonna, suggesting it must be Raphael.
The different results of the two AI programs have caused confusion in the art world, where valuations can fluctuate by millions of dollars depending on the authentication of a work by a master.
“Adoration of the Kings,” believed to have been painted by an artist related to Rembrandt, will be auctioned in 2021 with an estimated price of 10,500 to 16,000 euros ($13,500 to $21,000). was.
It was later confirmed to be by Rembrandt himself, and was sold at Sotheby's in December for £10.9 million (US$14 million).
According to Joe Lawson Tancred, author of the upcoming book AI and the Art Market, Art Recognition uses AI that excels at “pattern recognition.”
That model can recognize the salient features used by a particular artist, provided it is given enough previous, verified examples to “learn” from.
However, Lawson-Tancred added that human judgment is still essential for authentication, as it has not been developed enough to keep track of context.
Other experts are more skeptical. “I think a lot will depend on the data that is fed into the AI system,” Carlo Milano of Callisto Fine Arts told the FT.
“For example, if a questionable catalog raisonné is used to enter data about an artist, the conclusions may be questionable.”
Milano acknowledges that while AI may help reduce the margin of error in identifying genuine masterpieces, it can never completely replace hands-on human experience.
Some conservators are skeptical about whether AI can account for factors such as dirty layers of varnish, wear, and damage.
Professor Hassan Ugair, director of the Center for Visual Computing at the University of Bradford, said AI is unlikely to replace human work in the art world, but it could be a useful tool for authentication.
“Like spectroscopy and dating techniques, AI can also be an important tool in our primary toolbox,” he says.
“The main shortcoming at the moment is the quality of input given to the AI attribution programs currently in use,” art historian Bendor Grosvenor told the FT.
“It is simply impossible to determine whether a painting is by Rubens or not by relying solely on low-quality images of less than half of his works.
“Human appraisers cannot be trusted to do so, and neither can computers,” he concluded.