Machine Learning Elevates Old Masters’ Art Authentication

AI in art, even Old Masters. At Tefaf Maastricht, Carina Popovici shows AI attributing a painting to a Renaissance German artist.

By Sunil Sonkar 3 Min Read
3 Min Read
Machine Learning Elevates Old Masters' Art Authentication

AI is making an impact in the art world, even entering traditional areas like the Old Masters trade. At the upcoming Tefaf art and antiques fair in Maastricht, Carina Popovici, CEO of Swiss-based AI company Art Recognition, will showcase how AI recently attributed a painting to a Renaissance German artist.

Founded five years ago, Art Recognition boasts an AI system offering precise and objective authenticity evaluations of artworks. With over 500 evaluations completed, including verifying disputed works like a 1889 self-portrait by Vincent van Gogh, the company is gaining traction in the art authentication domain.

In art, figuring out who made a piece is really important. It can make a big difference in how much it is worth and help with academic studies. Recent instances, such as “The Adoration of the Kings” initially estimated as a “Circle of Rembrandt” but later attributed to Rembrandt himself, exemplify this impact, fetching a staggering £10.9mn at auction.

Advertisement

AI is good at noticing patterns, which helps it spot unique traits of artists when it sees enough examples. However, while AI can spot deviations from an artist’s style, it struggles with contextual understanding, requiring human intervention for nuanced assessments.

Even though AI is getting better, many art experts still have doubts. They worry that AI might not consider important things like varnish or damage, which are crucial for authenticating artworks, according to conservators.

The argument about the de Brécy Tondo painting, said to be by Raphael, shows both good and bad sides of AI authentication. Different AI programs give different answers, which we can see from the arguments of various research groups.

How well AI can authenticate art depends a lot on the quality and variety of the data it’s trained on. Popovici says it is crucial to show AI lots of different examples to avoid biased judgments.

While AI offers promise, it is not without limitations. Art historian Bendor Grosvenor stresses the need for better inputs and cautions against overreliance on AI for attributions.

Share This Article