Spratt, Emily L.:

The Kunstwollen of the Machine-Learned Image: Reflections on Riegl’s Legacy and the Ethics of Predictive Image-Based AI for Historic Preservation

The recent use of deep learning techniques in artificial intelligence has driven developments in computer-based visual analysis and production of digital images of art, architecture, and cultural heritage that are implicitly reviving the problematic premise that there are inherently predictable aspects of form that may be qualitatively interpreted by way of quantitative measurement. While the notion of artistic progression, or even the theory of artistic evolutionism, has captivated an audience that is today largely located outside the discipline of art history, this approach to art raises a host of ethical issues that scholars—particularly in the years following the Second World War—well addressed in their critiques of formalism. In this article, focus is placed on the legacy of Alois Riegl’s theory of Kunstwollen, which he expounded upon in his seminal book, Die Spätrömische Kunst-Industrie, as a concept approaching artistic volition on the level of form. Through consideration of the historic preservation of medieval architecture, its relationship to the burgeoning use of computer vision technologies in the visual arts is explored. If a theory of Kunstwollen does apply to the machine-learned image, the question arises, ''What are the ethical consequences of its presence and the implications for theories of the development of art on a global scale?'' Furthermore, although AI offers the possibility of digital reconstitutions in every capacity, what space is left for the singular art or archaeological fragment when its digital existence is wedded to larger image constellations that will its completion and connection to like forms?

Emily L. Spratt is an art historian, art technologist, and strategic advisor based in the Data Science Institute at Columbia University, where she has an appointment as a postdoctoral research fellow with co-sponsorship from the Departments of Historic Preservation and Computer Science. She is currently researching the development of AI-enhanced technologies for the analysis, generation, and curation of art and architecture, and the ethics surrounding this subject. Emily did her doctoral studies in the Department of Art and Archaeology at Princeton University on Byzantine and Renaissance art, and also holds degrees from the University of California, Los Angeles, and Cornell University. With much experience in the cultural heritage sector and the museum world, Emily also has taught in the Department of Art History and the program in cultural heritage and preservation studies at Rutgers University. Emily has been a consultant for The Frick Collection and Art Reference Library, was the former strategic advisor for Artory, the blockchain-based art market company, is the ethics advisor for Iconem, and is involved in sourcing, evaluating, and advising machine learning-related startup companies.