Van Eyck Attribution Questioned: AI Challenges Authenticity of St. Francis Paintings
Is the Golden Age of Art Authentication Over? AI Challenges Van Eyck Attribution
The art world is abuzz with a new challenge to traditional methods of attribution. Two near-identical 15th-century paintings of Saint Francis of Assisi Receiving the Stigmata, housed in the Galleria Sabauda in Turin and the Philadelphia Museum of Art, are facing renewed scrutiny. The question? Whether they were actually painted by Jan van Eyck, one of the most celebrated Early Netherlandish painters.
The Rise of AI in Art Forensics
A Swiss firm, Art Recognition, utilizing artificial intelligence, has cast doubt on the long-held attribution. Their analysis, conducted in collaboration with Tilburg University in the Netherlands, suggests that neither painting possesses the hallmarks of Van Eyck’s distinctive brushwork. The AI, trained on authenticated Van Eyck works, flagged the Philadelphia panel with a 91% probability of not being by the master and the Turin version at 86%.
This isn’t a new frontier for Art Recognition. The company gained attention in 2021 by questioning the authenticity of Rubens’ Samson and Dalila, a claim disputed by many experts. However, they also claim success in identifying forty counterfeit paintings on eBay in 2024, highlighting the potential of AI to combat art fraud.
How Does the AI Work?
Art Recognition’s method relies on ultra-high-resolution photography, breaking down images into thousands of fragments. The AI then extracts recurring characteristics and compares them to a database of confirmed works. When applied to Van Eyck’s undisputed Portrait of the Arnolfini Couple, the AI correctly identified it with an 89% probability.
A Divided Art Historical Community
The AI’s findings aren’t universally accepted. Some art historians, like Till-Holger Borchert, director of the Suermondt-Ludwig Museum in Aachen, see the results as supporting the theory that the two Saint Francis panels are workshop creations, rather than directly from Van Eyck’s hand.
However, others raise concerns about the methodology. Maximiliaan Martens, a professor at Ghent University, points out that Van Eyck was known for a smooth, almost imperceptible brushstroke, achieved through glazing techniques. This makes identifying a “touch” for the AI to analyze particularly difficult. The limited number of definitively attributed Van Eyck paintings – fewer than twenty – creates a small dataset for training the AI, potentially impacting its reliability.
Beyond Brushstrokes: The Challenges of Digital Authentication
Martens also highlights the impact of material factors. The Philadelphia version is painted on a rare vélin support, which alters the surface and could mislead the AI. Centuries of varnish oxidation, craquelure, retouching, and restoration further complicate the analysis, potentially causing the AI to mistake aging and alteration for stylistic traits.
The Future of Art Authentication: A Hybrid Approach?
The Van Eyck debate underscores a growing trend: the integration of AI into art historical research. While AI offers powerful new tools for analysis, it’s unlikely to replace traditional connoisseurship entirely. The most promising path forward appears to be a hybrid approach, combining the objective data provided by AI with the nuanced judgment of experienced art historians.
Will AI Democratize Art Expertise?
Currently, art authentication relies heavily on the expertise of a relatively small number of specialists. AI could potentially democratize access to authentication tools, allowing museums and collectors with limited resources to conduct preliminary assessments. However, the cost of the technology and the need for specialized knowledge to interpret the results remain significant barriers.
The Expanding Role of Data in Art History
The use of AI in art authentication is part of a broader trend toward data-driven art history. Researchers are increasingly using digital tools to analyze large datasets of images, texts, and archival materials, uncovering new insights into artistic practices and historical contexts. This shift promises to transform the field, offering new avenues for research, and interpretation.
FAQ
Q: Can AI definitively prove a painting is not by a specific artist?
A: Not definitively. AI provides a probability assessment based on the data it has been trained on. It can raise doubts, but human expertise is still needed for final judgment.
Q: What are the limitations of using AI for art authentication?
A: Limited datasets of authenticated works, the difficulty of analyzing artists with subtle techniques, and the impact of material factors like varnish and restoration can all affect the accuracy of AI analysis.
Q: Is AI only useful for identifying forgeries?
A: No. AI can also help to refine our understanding of an artist’s style, identify workshop contributions, and trace the provenance of artworks.
Did you know? Jan van Eyck is credited with perfecting oil painting techniques, allowing for unprecedented levels of detail and realism.
Pro Tip: When researching art authentication, always consult multiple sources and consider the perspectives of different experts.
What are your thoughts on the role of AI in art authentication? Share your opinions in the comments below!