The Generalisation Trap

Domain-Specific Knowledge, Human Expertise and the Future of AI — A JMCE School of Diplomacy Lecture

How is artificial intelligence transforming the nature of professional expertise, and what role will human knowledge play in the future of work? As AI increasingly reshapes knowledge-intensive professions, this talk explores the relationship between technological innovation, specialised human expertise and the changing boundaries of professional labour. Drawing on the emerging field of computational art history, it examines how traditional humanities disciplines can contribute to the development of more sophisticated, context-sensitive machine learning systems. Focusing on the concept of generalisation trap, the limitations of AI models when confronted with domain-specific knowledge and historical complexity, the talk investigates why human expertise remains essential, not merely as a complement to AI, but as an active force shaping its development.

Venue and Information The event will be held in Bush House South East Wing , Room BH(SE)1.02. F or checking in, Guests should proceed to Bush House South East Wing reception, situated at 300 Strand, WC2R 1AE. An informal reception with drinks will follow. We encourage all attendees to stay, enjoy a drink, and network. Please note, King's events are free, which means we routinely overbook to allow for no-shows and avoid empty seats. Admission is on a first come, first served basis, so please arrive in good time to avoid disappointment. We will not be able to admit those without tickets. There may be some photography at this event for archival and promotional purposes. If you would prefer not to be photographed, please let a member of staff or the photographer know on arrival.

Speaker Alessandra Conti — Art Historian Alessandra Conti is an art historian and Lead Art Historian at Art Recognition AG in Zurich. Her work explores the intersection of art history, computer vision and artificial intelligence, with a particular focus on how art-historical knowledge can inform and shape AI-based approaches to the analysis of artworks. Her research interests include the detection of non-authentic restoration and repainting, as well as the iconography of dress in Early Modern art. She co-developed the methodology and technology for the AI-based analysis of restored and repainted areas in artworks, published in Kunstgeschichte: Open Peer Reviewed Journal (2026). She is also co-author of Art-Historically Grounded Multimodal Pretraining for Art Authentication, presented at the 4th AI for Visual Arts Workshop (2026) and forthcoming in a Springer publication. Her research has been presented at institutions including Ca’ Foscari University of Venice and the University of Bern.

About the JMCE School of Diplomacy The JMCE School of Diplomacy is a series of lectures given by academics, diplomats or accomplished practitioners on the future of the relations between the EU and the UK in the financial, legal and political…