Spanish Renaissance master El Greco is often considered one of the greatest painters of all time, and many of his artworks are displayed in galleries around the world. His painting The Baptism of Christ is generally believed by art historians to have been unfinished at the time of his death in 1614 and completed by his son, Jorge Manuel. But new research using AI suggests that the artist may have played more of a role in completing the work than previously thought.
El Greco, like many well-known artists, used workshops of skilled apprentices to help him paint his commissioned works. This presents a challenge for modern AI technology to determine who painted which part of a masterpiece. That’s because numerous artists can mean a range of styles on one canvas, and normally, to train AI models, you need a reference library of known examples to work from.
But scientists got around this by developing a new technique called PATCH (pairwise assignment training for classifying heterogeneity), as they write in their paper published in the journal Science Advances.
A new tool for art sleuths
PATCH works by first scanning the surface of a painting with a high-resolution 3D imaging technique to create highly detailed maps of the paint texture. It captures the tiny ridges and valleys left behind by an artist’s brush.
Then the researchers used a type of AI called a convolutional neural network to compare two one-centimeter square patches. If it can tell them apart, they likely came from different artists. If it struggles to distinguish them, they were more likely created by the same artist. Depending on the size of the scanned area, PATCH may compare hundreds of patches.
“PATCH outperforms statistical and unsupervised machine learning methods in this complex pairwise comparison problem lacking ‘ground truth’ data,” commented the researchers in their paper.
The team tested their technology on twenty-five paintings created by nine students to prove the AI could accurately tell different hands apart.
Then they applied PATCH to The Baptism of Christ and another El Greco painting, Christ on the Cross with Landscape (widely agreed to be largely completed by El Greco himself). Overall, the analysis revealed a generally consistent hand at work with both paintings.
A valuable tool
The researchers do not claim their PATCH work is conclusive. The system has its limitations, as damage and age can throw it off, for example. And if El Greco’s students were skilled at mimicking his work, AI might not be able to tell their hands apart. Nonetheless, the study authors believe their research is a valuable contribution to the efforts to solve this mystery.
“Our findings regarding one of the works potentially challenge previous studies that assert that a considerable portion of the painting was completed by workshop members after El Greco’s death,” said the study authors.
Written for you by our author Paul Arnold, edited by Gaby Clark, —this article is the result of careful human work. We rely on readers like you to keep independent science journalism alive.
If this reporting matters to you, please consider a donation (especially monthly). You’ll get an ad-free account as a thank-you.
Publication details
Andrew Van Horn et al, PATCH: A deep learning method to assess heterogeneity of artistic practice in historical paintings, Science Advances (2026). DOI: 10.1126/sciadv.aea0485
© 2026
The content is provided for information purposes only.