The painting “Samson and Delilah” for decades was considered the pearl of the National Gallery in London and even longer caused controversy among art critics: some of them refused to believe that the painting belonged to Rubens. The program, using artificial intelligence technologies, confirmed the suspicions of skeptics: AI considered this picture to be almost certainly a fake.
The truth about “Samson and Delilah” was told by the Swiss company Art Recognition, which specializes in using AI to determine the authenticity of works of art. There is nothing super complicated in this process. AI is “taught” on a large array of other works by the same artist, so that in the end it can reveal the peculiarities of writing down to the nature of brush strokes. For a more accurate analysis, the machine learning canvas is divided into many small sections, each of which is analyzed by AI separately. Thus, AI studies hundreds of works of a particular artist and can with a high degree of probability detect a fake.
Moreover, AI determines the characteristic style of a particular artist, even if he has noticeably changed his writing style throughout his life.
Art Recognition cooperates with the Netherlands Tilburg University and has already analyzed about 400 paintings in this way.
In Rubens’s case, AI was trained in 148 of his jobs. As a result, when analyzing “Samson and Delilah” AI came to the conclusion that the work with a probability of 91.8% does not belong to the brush of the great Dutchman. “The result was amazing. We repeated the experiment to make sure we didn’t make a mistake, but the result was always the same. Each … individual square turned out to be a fake with a probability of more than 90%, ”says Karina Popovichi, co-founder of Art Recognition.
For art critics, this did not come as a big surprise – disputes about the authenticity of the painting have been going on for decades. The National Gallery bought Samson and Delilah in 1980 at Christie’s for a record £ 2.5 million at the time.
The painting was discovered – or, as we can now say with some degree of probability, painted – in 1929.
At first it was attributed to another Dutch painter of the 17th century – Gerrit van Honthorst, only then the German historian, art critic and renowned expert on Rubens’ work Ludwig Burchard said that Samson and Delilah belonged to the brush of this artist. However, after the death of Mr. Burchard, art critics discovered that not all of his statements about the authenticity of the paintings are true.
Art historians agree that Rubens really wrote “Samson and Delilah” in 1608–09 by order of the burgomaster of Antwerp Nicholas Rockox, and already around 1640, after the death of Rockox, the painting disappeared.
According to skeptics, her further fate is unknown, and the work exhibited in the National Gallery does not belong to Rubens.
Here are just some of the arguments that various art historians have put forward in favor of the fact that the painting “Samson and Delilah” is a fake: too rough and firm manner of painting in comparison with other works by Rubens; a composition that this artist would hardly have chosen; an uncharacteristic choice of colors and paints for him, and even the artist’s insufficient knowledge of anatomy, which, in their opinion, can be seen in this work. As noted by the American artist and art critic Jane Park, this work “was written (by a person – “B”) with a post-impressionist way of thinking, although it was conceived as a baroque painting. “
Now, AI has added to these arguments. “The implications of this new AI authentication method could be revolutionary. Deprived of human subjectivity, emotions and commercial interests, software has cold objectivity and scientific accuracy. At the beginning of the 20th century, many dubious works were attributed to Rubens … Now there is an urgent need for more reliable methods of determination, ”art historian Kasia Pisarek believes.
The National Gallery, which has so far rejected doubts about the authenticity of “Samson and Delilah”, now said that they “expect” the publication of the study in its entirety, and so far did not consider it possible to comment on it.
This is not the first time artificial intelligence has been applied to works of art: for example, scientists have already created AI, which can predict how people will react to a particular work of art, as well as AI, which provides a quick estimate of the value of works of art.
.