Staring vaguely at something unseen to her right, the black-haired woman strikingly resembles the person depicted in a Pierre-Auguste Renoir painting. Gabrielwhich Sotheby’s recently estimated at £100,000 to £150,000.
However, art experts disagree on whether the work, which is owned by a private Swiss collector, is the real one. Now, artificial intelligence has stepped in to help settle the dispute, and the computer has deemed it likely that Renoir could be real.
Artificial intelligence is increasingly being used to help decide whether valuable works of art are real or fake. Earlier this month, Art Recognition, the Swiss company that developed the technology, announced that it had concluded that Switzerland’s only Titian — a work called Evening Landscape with Couple, held by the Kunsthaus Zürich — may not have been painted by the 16th-century Venetian artist.
However, art experts have warned that the AI is only as good as the paintings it rehearses. If it is fake, or contains areas that have been touched up, this can create more uncertainty.
The art recognition about Renoir, titled Portrait de femme (Gabrielle), is approached after the Wildenstein Plattner Institute—one of two institutes that publish a comprehensive list of all known artworks by Renoir, known as Catalogs – He refused to include it in his list.
The company used photographic reproductions of 206 original paintings by the French Impressionist artist to teach its algorithm about his style, which for human observers is characterized by broken brush strokes and bold combinations of complementary colors. To increase accuracy, it also divided the images into smaller patches and displayed them to the algorithm, as well as training them on a selection of paintings by artists in a similar style who were active around the same time as Renoir.
Based on this assessment, it is concluded that there is an 80.58% chance that Portrait de Femme (Gabriel) Drawing by Renoir.
Karina Popovici, CEO of Art Recognition, believes that this ability to quantify uncertainty is important. Speaking at a meeting on the forensic use of technology in the art trade at the Art Loss Register in London on Monday, she said: “Art owners are often told by experts that their ‘impression’ or ‘hunch’ is that a painting is truthful or not, which may It’s very frustrating. They really appreciate the fact that we’re being more punctual.”
Encouraged by this finding, the author of the painting approached another Parisian connoisseur, GP.F Dauberville & Archives Bernheim-Jeune, who is publishing its own collection indexes Renoir’s work. After ordering a scientific analysis of the pigments in the painting, they also concluded that it was a real Renoir.
Dr Bendor Grosvenor, art historian and presenter of BBC Four’s Britain’s Lost Masterpieces, expressed concern that such techniques could underestimate the input of experts in assessing the authenticity of a work of art.
“So far, the methods used to ‘train’ the AI programs, and the fact that they say they can judge attribution solely from an iPhone image, are inconsequential,” he said.
“The technology is particularly weak in its inability to take into account the condition of the board — many older motherboards are damaged and marred by layers of dirt and overpaint, making it difficult to tell what is original and what is not without a forensic examination.
“If any human art appraiser offered to give out a ‘certificate of authenticity’ costing thousands of dollars based on no more than an iPhone photo and partial knowledge of the artist’s work, it would be an object of ridicule.”
Popovici agreed that the quality of the training data set is vital, and said they go to great lengths to ensure they only use images of original artwork. So far, they’ve trained their AI to recognize about 300 artists, including most of the French Impressionists and Old Testament painters.
“We understand that experts may feel threatened by this technology, but we’re not trying to get them out of the way,” Popovici said.
“We really want to give them the possibility to use this system to help them come to a decision, maybe in cases where they’re not quite sure. But for that to happen, they need to be open to this technology.”
Julian Radcliffe, head of the Art Loss Registry, which maintains the world’s largest private database of stolen art, antiques and collectibles, said: “Artificial intelligence has a growing role in helping to document art but it must be allied with the expertise of experts specializing in art, and established sciences such as tint analysis, and source research.
Its advantage lies in its ability to provide yes/no answers, eg pattern or matching analysis, and continuous improvement, but its work must be interpreted by the human who must have identified the correct question.
“The quest for absolute certainty in authentication has not been achieved, and may never be reached – but we are getting closer.”