A group of scientists at Case Western Reserve University in Cleveland, Ohio, has built a new machine-learning program that might help art historians determine how many artists’ hands contributed to the creation of centuries-old artworks. The dozen researchers who published the paper in Science Advances range from physicists and computer scientists to art historians and anthropologists.
The AI model is called PATCH, which stands for pairwise assignment training for classifying heterogeneity. It works by comparing 1-centimeter-square “patches” of artworks that are known to have been painted by an individual artist (rather than a group of artists, or a workshop, as was common during the early modern period). The consistency in brushwork and paint texture found in these tiny segments can then be compared to another artwork with more questionable background to indicate if it was made by the same individual artist, a different individual artist, or perhaps a group.
The paper in question focused on the Spanish Renaissance painter El Greco. The researchers used PATCH to compare El Greco’s Christ on the Cross, which “has been entirely attributed to El Greco himself,” and The Baptism of Christ, which “has long been thought to feature the work of El Greco and at least one other artist.” The Baptism of Christ is generally believed to have been completed by El Greco’s son and other apprentices in his workshop in the decade after the artist died. But, the machine-learning model comparing scanned areas of each painting indicates that The Baptism of Christ might, in fact, have been painted primarily by just El Greco, with little or no contributions by his workshop.
The paper’s researchers don’t claim that AI models like PATCH can definitely answer technical questions related to art history scholarship of this era. (In fact, a recent announcement that artificial intelligence had proven that a long-contested artwork had been painted by Raphael quickly came under fire by museum professionals.) Rather, PATCH “has the capability to make a substantial contribution to research as a complement to existing art historical methods.”
