Digital Humanities and AI for Cultural Heritage

The questions that concern me here are less about the deployment of any particular technique and more about how the field is changing as a whole. Digital systems now sit between scholars and the sources they study — in transcription, in subject indexing, in segmentation, increasingly in summarisation — and the choices made by those systems are not neutral. They privilege some kinds of source, marginalise others, and embed assumptions about what counts as a “page”, a “paragraph”, or a “topic” that historians have spent decades unpacking.

My contribution to this conversation has two registers. The first is synthetic and diagnostic: I have co-authored two State of the Field articles surveying the state of digital methods in historical research, one for the broader field and one specifically for legal history. Both ask not only what has been done but what the field’s principal methodological tensions and blind spots are.

The second register is practical and infrastructural. The questions I ask theoretically — about who owns the tools, how ground truth is generated and attributed, whether AI-driven transcription and metadata creation can be trusted and for what purposes — are also the questions I encounter daily in the HAICu project, in the HISMET project, and in my work as Chair of the Board of READ-COOP. There is no clean separation between the analytic and the institutional here; the cooperative model of READ-COOP and the archive-side focus of HISMET are both, in part, answers to the critical questions this research area poses.

I write about these matters for History, DSH, Digital Humanities Quarterly, and the International Journal of Digital Humanities, and present regularly at DHBenelux, DHNB, and DH conferences.

Related projects