The following abstracts and biographies provide an overview of the conference Archival Intelligence: AI × Archives × Museums.
Abstracts | Biographies |
| Dominik Bönisch Alpári | University of Applied Sciences Düsseldorf |
| Inside the Toolbox: Exploring AI-Driven Prototypes in Archives and Museums The presentation offers insights into pragmatic approaches for making museum collections and cultural archives, which comprise diverse images as well as audiovisual materials embedded in complex linguistic contexts, more accessible and explorable in new ways. The talk introduces three specific prototypes developed in projects that employ search algorithms, generative AI, and Custom GPT models. It aims to encourage everyone to experiment with the available tools. To support this, practical accounts of the application design process and brief tutorials will be shared, creating opportunities to integrate one’s own experiences and exchange lessons learned. | Dominik Bönisch Alpári studied Cultural Studies at the University of Hildesheim and the Moholy-Nagy University of Art and Design in Budapest. He is currently working as a research associate at the University of Applied Sciences Düsseldorf (MIREVI), focusing on the exploration of intermedia collections using artificial intelligence. Since 2024, he has been managing the project LINKed, investigating how audiovisual archives can be connected with AI. Until 2023, he headed Training the Archive at the Ludwig Forum Aachen, where he applied AI in curating. His research interests focus on the impact of AI on museum collection and archival practices |
| Giovanni Colavizza | University of Copenhagen |
Using AI to Broaden Access to Historical Archives Artificial Intelligence (AI) is crucial in supporting archival processes and records management decisions, including for increasingly digitized historical archives. The scale and complexity of historical archives contribute to make AI ever more relevant for improving their organization and broadening their access. In this talk I will discuss recent developments at the intersection of archives and AI, as well as highlight some of the challenges still lying ahead of us. I will also discuss recent work to exemplify the current state of AI applied to historical archives. Finally, I will suggest future research directions in respect thereof. | Giovanni Colavizza is Professor and Head of the Centre for Digital and Computational Humanities at the University of Copenhagen, Denmark. He is also an Associate Professor of Computer Science at the University of Bologna, Italy, and the CTO and co-founder of Odoma, a Swiss-based studio. Colavizza has a background in Computer Science and History and is specialized in AI applications in the GLAM sector (Galleries, Libraries, Archives, Museums). |
| Robert Erdmann | University of Amsterdam |
Artificial Intelligence Across the Museum Workflow: Conservation, Collections, Curation When embedded in carefully designed custom software, recent advances in AI can provide dramatic benefits across cultural heritage workflows, including advanced imaging, scientific analysis, condition assessment, visualizing evidence to aid attribution, collection management, and public outreach. Live demos of the author’s projects provide examples of each, from the creation and analysis of a 717-gigapixel image of Rembrandt’s Night Watch to a real-time app that searches more than 10 million cultural-heritage images per second. Together, these and several other applications showcase the responsible use of AI to help the world access, preserve, and understand its cultural heritage. | Robert Erdmann (Ph.D., University of Arizona, 2006) founded a scientific software company, did metal casting research at Sandia National Laboratories, and was Professor of Materials Science and Applied Mathematics at the University of Arizona. In 2014 he moved to Amsterdam to unite materials, computer, and imaging sciences for cultural heritage, serving 2014–2024 as Senior Scientist at the Rijksmuseum. Since 2014 he has been Full Professor of Conservation Science at the University of Amsterdam. He received the 2017 Europa Nostra Grand Prix for the Bosch Research and Conservation Project and since 2025 has been a Research Fellow in AI at the Museum of Decorative Arts in Paris. |
| Ralph Ewerth | Philipps University Marburg, TIB Leibniz Information Centre for Science and Technology, Hannover |
Using the TIB AV Analytics Platform for Analyzing Disinformation Patterns in News Videos Automatic video analysis enables applications in many disciplines including film and media studies, communication science, and education. In this talk, I will present the web-based video analysis platform entitled TIB AV-Analytics (TIB-AV-A) which integrates state-of-the-art approaches from the fields of computer vision, audio analysis, and natural language processing for relevant video analysis tasks. To facilitate future extensions and to ensure interoperability with existing tools, a plugin structure with appropriate interfaces and import-export functions is utilized. TIB-AV-A leverages modern web technologies to provide users with an interactive web interface that enables manual annotation and provides access to powerful deep learning methods without a requirement for specific hardware. As a use case for TIB-AV-A, I will present results from the research project “FakeNarratives” (funded by German Ministry of Education and Research) on the analysis of news videos. Here, we first investigated whether the narrative strategies for broadcast and “alternative” media differ, and second, whether there are typical narrative patterns for conveying disinformation. An example result from this project is our approach for the identification of speaker roles and situations in news, which is a prerequisite for the analysis of narrative patterns. | Ralph Ewerth has been Professor of Multimodal Modeling and Machine Learning at Philipps University Marburg since April 2025 and is a member of the Hessian Center for Artificial Intelligence “hessian.AI”. Previously, he was Professor of Visual Analytics at Leibniz University Hannover from 2015 to 2025. Since 2015, he has led a research group at TIB (German National Library of Science and Technology), the Leibniz Information Centre for Science and Technology, and is a member of the L3S Research Center in Hannover. Prof. Ewerth has published more than 150 scientific articles, particularly on multimodal search, multimodal data analysis, computer vision, digital libraries, and technology-enhanced (human) learning. |
| Jasmijn Van Gorp | Utrecht University |
Repeated TV: Methods to Study Archived Television with AI Television as a medium is characterised by temporal patterns of reuse and repetition. As William Uricchio notes, this “recombinatory practice” is particularly difficult to recover from television archives. In this presentation, I critically examine several methods, drawing on AI-generated speech-recognition files and “manual” image recognition, to identify forms of repetition in archived television. Taken together, these methods demonstrate how archived television is the epitome of a recombinatory practice in which repetition is intertwined with archival reordering processes. The presentation also shows the extent to which copyrights and restricted access significantly shape current research on AI and archives. | Jasmijn Van Gorp is a Senior Lecturer in Audiovisual Heritage and Digital Culture at Utrecht University, the Netherlands. She is the co-founder of the CLARIAH Media Suite, an integrated digital research infrastructure for audiovisual heritage in the Netherlands. Her research interests include AI techniques (e.g. ASR), digital humanities, creative methodologies, and user-centred approaches to audiovisual archives. She is currently working on the project Re-Frame: Audiovisual Data and AI in Journalism Practice. She has published widely at the intersection of digital humanities and audiovisual archives, including in Digital Humanities Quarterly and the Historical Journal of Film, Radio and Television. |
| Adelheid Heftberger | German Federal Archives (Bundesarchiv), Berlin |
AI Projects at the Bundesarchiv – First Steps and Initial Findings In my presentation, I will introduce two current AI projects at the Federal Archives against the backdrop of internal discussions on the topic of AI and discuss our experiences and challenges. I will focus on considerations regarding our audiovisual holdings. I will also place the Federal Archives’ first steps in a broader context of similar discussions within the international film archive community. Finally, I would like to venture an outlook on future applications (or lack thereof) of AI technology at the Federal Archives. | Adelheid Heftberger is the Deputy Head of the Department for Audiovisual Media at the German Federal Archives (Bundesarchiv) in Berlin. Previously, she worked as a researcher, curator, and archivist at the Brandenburg Center for Media Studies in Potsdam and at the Austrian Film Museum in Vienna. She holds a PhD in Slavic Studies and a Master’s degree in Comparative Literature from the Universities of Innsbruck and Vienna. In 2016, she completed an M.A. in Library and Information Science at Humboldt University of Berlin. Dr. Heftberger currently chairs the FIAF Cataloguing and Documentation Commission and is an active advocate of the Open Science movement. |
| Andreas Kohlbecker | ZKM |Karlsruhe |
Experiments, Prototypes, and Perspectives – AI-Supported Access to the ZKM Archive Recent advances in AI, powered by open-source tools, enable rapid prototyping. At the ZKM archive, we have developed experimental workflows and prototypes that use AI to improve the access to manuscripts containing hand-drawn electronic circuits and textual funds, as well as conceptual perspectives for time-based media, including cases where access is legally restricted. This talk presents selected results from these efforts, identifies limitations in the AI tool ecosystem, especially regarding the multimodal analysis of large time-based media collections in GLAM, and outlines perspectives for responsible and scalable approaches to unlocking archival knowledge. | Andreas Kohlbecker is Digital Manager at the ZKM | Karlsruhe, leading the development of state-of-the-art data infrastructure and making archives and collections accessible to the public and within scientific data spaces. He holds a diploma in biology from the University of Karlsruhe. As a research assistant, he worked at the European Media Laboratory in Heidelberg on modeling biochemical pathways and at the University of Bayreuth on software development for the Biologische Bundesanstalt für Land- und Forstwirtschaft in Berlin. From 2007 to 2022, he was a research associate in the Biodiversity Informatics working group at the Botanic Garden and Botanical Museum Berlin-Dahlem, Berlin. Since the late 1990s, he has been active in media art, a member of the art collective Versfabrik, and co-founder of Raumfühler, an artistic exploration of electromagnetic fields. |
| Bárbara Romero Ferrón | Leuphana University of Lüneburg |
The Epistemologies of Provenance: Textual Practices, Humans, and Artificial Intelligence Working with provenance often means engaging with provenance records that can be understood as a particular “literary genre:” narrative, formulaic, fragmentary, and frequently ambiguous. The epistemology embedded in these narratives, together with other key sources such as exhibition histories, poses a significant challenge: dense human knowledge compressed into concise formulations. Within the Modern Migrants project at the Provenance Lab, we have been designing and exploring methodologies that use NLP and LLMs, together with art-historical and provenance expertise, to transform these records into structured, reusable data while preserving their interpretive richness and nuances. | Bárbara Romero Ferrón is a Research Associate at the Provenance Lab, Leuphana University of Lüneburg, and an affiliated senior researcher at iArtHisLab, University of Málaga. She holds a BA and MA in Art History from the University of Málaga and a PhD in Art History from the CulturePlex Lab, Western University (Canada). Her work lies at the intersection of provenance, exhibition, and curatorial studies, with complex systems theory and network analysis, and is grounded in feminist and queer intersectional perspectives. Beyond art history, her work also focuses on broader cultural phenomena, including women’s literature and correspondence, convent writing, and models of female sanctity. She is also involved in a range of research projects and exhibition initiatives with institutions such as the Spanish National Research Council (CSIC), the Complutense University of Madrid, the Getty Research Institute, and other national and international museums and research institutes. |
| Heiko Schuldt | University of Basel |
Ask not what your archive or museum can do for you – ask what vitrivr can do for your search Over the past few years, the proliferation of powerful AI tools has greatly advanced the analysis of large text and multimedia collections. However, searching such collections has not yet reached its full potential, especially when considering historic content or audiovisual data. In this talk, I will introduce vitrivr, a full-stack, open-source multimedia retrieval system that addresses these limitations. Additionally, I will present recent applications of vitrivr in the cultural heritage context. | Heiko Schuldt is full professor of Computer Science at the University of Basel, Switzerland, where he leads the Databases and Information Systems (DBIS) research group. He holds a PhD in Computer Science from ETH Zurich and a diploma from the University of Karlsruhe. His research interests include large-scale Distributed Information Systems, Multimedia Retrieval, as well as Big Data Processing and Analytics. |
| Christiane Sibille | ETH Zurich |
Transform – Evaluate – (Re-)Use: Collections as Data in Practice Since 2021, the Digital Scholarship Services office of the collections and archives at the ETH Library has been developing projects and services at the intersection of digitized collections, infrastructure, and research. The idea of “collections as data” and the use of machine learning technologies have been central to this work from the very beginning. These projects currently focus on workflows for transforming and enriching digitized collections, ranging from handwritten transcripts to richly illustrated journals. In this context, we have developed a four-stage conceptual framework that helps us identify different levels of application for machine learning in archives. In my presentation, I will introduce this framework and discuss the learnings from our projects, focusing on the role of interdisciplinary methods in establishing sustainable workflows and services. | Christiane Sibille studied history and musicology in Heidelberg and now works at the (digital) interfaces between research, archives, and libraries. Her professional career has included positions at the Universities of Heidelberg and Basel, the Zurich University of the Arts, the Diplomatic Documents of Switzerland (Dodis) research center, and the Metagrid.ch project. Since June 2021, Dr. Christiane Sibille has been responsible for the Digital Scholarship Services department in the Collections and Archives section of the ETH Library. Additionally, she regularly teaches digital topics at the University of Basel and is president of the History and Computing Association. |
| Kim Voss | German Broadcasting Archive |
Unlocking GDR Broadcasting History with AI: A Journey Through Archival Automation At the German Broadcasting Archive a dedicated Automation Team was established in 2022 with the goal to bring AI and automated workflows into archival processes. Since then, the team has experimented with Large Language Models and other AI tools for transcription and indexing. Based on these experiences, they set up projects to integrate the tools into daily documentation workflows. The team is currently finalizing a web application that enhances AI analysis of GDR broadcasting records and provides tools for validating outcomes before database integration. This presentation will detail the challenges encountered in applying AI to often fragile and complex historical recordings and the iterative development process that led to a specialized web application. | Kim Voss is a data and information specialist at the German Broadcasting Archive. In her current projects, she is integrating automation and AI in audio documentation workflows. Her expertise is at the intersection of archival processes, data analysis, programming and technologies. She has a background in media und cultural theory and worked as project and communications manager in the fields of art, culture and digital media. |