Duke Researchers Awarded Grant to Reveal Hidden Histories of Artworks Through AI and Imaging

An interdisciplinary team of Duke faculty is among the international recipients of a new grant from the Schmidt Sciences Humanities and Artificial Intelligence Virtual Institute (HAVI), a program designed to bridge the gap between AI technology and humanistic inquiry. 

Led by Martin Fischer, Research Professor of Chemistry and Physics; Shira Faigenbaum-Golovin, Assistant Professor of Mathematics from Bar-Ilan University (formerly of Duke); and John K. Delaney, from the National Gallery of Art in Washington, the project, “From Molecules to Masterpieces,” will use advanced molecular imaging, mathematical modeling and machine learning to reveal the hidden histories buried beneath the surface of paintings and cultural artifacts. The project aims to allow researchers to virtually “rewind time” to understand how artworks once looked, how they were created and how they’ve changed through the centuries, offering a new window into the creativity and history preserved in each masterpiece.

This approach could benefit museum conservators, art historians and scientists alike. It will enable scholars to understand artists’ methods and materials, guide restoration efforts with scientific precision, and contribute to questions of authorship by identifying stylistic and material patterns unique to different creators or periods.

Beyond the arts, the machine learning methods developed by the team to analyze complex, multimodal data from paintings may have broader applications in fields such as biological imaging, where similar challenges arise in fusing and interpreting varied data sources.

In addition to Fischer, the Duke team includes (listed alphabetically) Ingrid Daubechies, James B. Duke Distinguished Professor Emeritus of Mathematics and Electrical and Computer Engineering; Roarke Horstmeyer, associate professor of Biomedical Engineering; Julia McHugh, from the Nasher Museum of Art; Edward Triplett, assistant professor of Art, Art History & Visual Studies; Warren Warren, James B. Duke Distinguished Professor of Chemistry, Physics, and Radiology; and Katherine Werwie, from the Nasher Museum of Art. 

Other collaborators on the team include Katrien Keune, Rijksmuseum (Netherlands); Narayan Khandekar, Harvard Art Museums; Roxanne Radpour, University of Delaware; Samantha Skelton, North Carolina Museum of Art; Erich Uffelman, Washington and Lee University; and Kristin Wustholz, William & Mary University.

Through HAVI, Schmidt Sciences has awarded $11 million to up to 23 research teams worldwide. The initiative encourages collaborations that apply AI to deepen cultural and historical understanding and use humanistic perspectives to shape how AI systems are designed and deployed. “From Molecules to Masterpieces” builds upon Duke’s tradition of cross-cutting research that unites science, engineering and the humanities. By linking AI with molecular imaging and art history, this collaboration stands at the forefront of a growing movement to use digital tools not to replace, but to enhance human understanding of cultural heritage.