DARIAH-IS / CLARIN-IS Day 2026
Reykjavík 26 November 2026 in Edda from 13.00 to 16.00
The event is open to all and free of charge
Research in the humanities has been shaped in recent years by the advent of artificial intelligence, in particular large language models. These tools are now part of everyday research practice, yet much of what is said about them is either promotional or dismissive. This afternoon workshop, jointly organised by the Centre for Digital Humanities and Arts / DARIAH-IS and CLARIN-IS, takes a different route: it looks closely at projects in which researchers have used AI, often alongside other digital methods, as one part of a research process – reading old newspapers, transcribing folklore recordings, describing photographic archives, interpreting media art.
The aim is to show how this work is actually done: what the tools made possible, where they fell short, what had to be checked by hand, and how the research questions themselves shifted along the way. The speakers come from libraries, research institutes, universities and museums in Iceland and Latvia, and their examples are offered as material for critical and practical reflection.
The afternoon ends with coffee and a Digital Humanities “Carpentry”, where PhD students and postdocs can bring projects or project ideas that draw on AI or other digital research methods and work through them with the speakers and other experts.
Agenda
13.00-13.10 Eiríkur Smári Sigurðarson: Introduction
13.10-14.00 Anda Baklāne: Multimodal AI in Art Collection Research: The Case of the Zenta Photo Explorer
14.00-14.20 Trausti Dagsson: Making sense of cultural data with AI on the side
14.20-14.45 Gundega Skela: AI in Museums: Emerging Practices and New Forms of Interpretation
14.45-15.05 Steinþór Steingrímsson: „What was playing at the movies?“ (And how OCR and small language models can help us create new research tools from old materials.)
15.05-16.00 Coffee and a Digital Humanities Carpentry
Digital Humanities Carpentry. Are you a PhD student or postdoc using, or considering, AI tools in your research – or working with other digital methods such as OCR, transcription, text analysis, mapping or databases? Bring your project, a sample of your material and your questions. In small groups organised by type of material (text, images, sound, archives and collections), you will discuss with experienced researchers what AI and other digital tools can realistically do for your project, where a simpler method might serve better, how to check the output, and what to document along the way. Please describe your project briefly when you register.
To participate in the carpentry, please register HERE.
Abstracts
Anda Baklāne, National Library of Latvia
Multimodal AI in Art Collection Research: The Case of the Zenta Photo Explorer
Advances in multimodal AI in the recent years have introduced new possibilities for exploring, describing, and analysing image collections. While earlier image recognition models relied primarily on visual patterns learned from training data, contemporary models combine visual and linguistic capabilities with extensive knowledge of art, history, and cultural contexts.
This presentation introduces Zenta Photo Explorer, an AI-powered application developed at the National Library of Latvia for exploring the photographic archive of Latvian photographer Zenta Dzividzinska. The application supports semantic image search, automated image tagging, and visual analysis of the collection through interactive charts. Using this project as a case study, the presentation will discuss the practical possibilities and limitations of multimodal AI in working with photographic archives, including the relationship between automated descriptions, curatorial knowledge, and research questions.
Trausti Dagsson, Árni Magnússon Institute for Icelandic Studies
Making sense of cultural data with AI on the side
Ísmús gives access to the folklore audio collection of the Árni Magnússon Institute for Icelandic Studies, together with databases of folk legends, music and vernacular poetry and information on the people connected to each recording. This talk discusses how speech recognition, Optical Music Recognition (OMR) and Optical Character Recognition (OCR) models have been used to work with these materials – making the data more legible and building stronger connections between entities in the database – with AI serving as a means of making sense of the collection rather than an end in itself.
Gundega Skela, University of Iceland
AI in Museums: Emerging Practices and New Forms of Interpretation
This talk surveys recent trends in the use of artificial intelligence in museums: the practices now emerging, the opportunities they offer and the questions they raise. It then introduces a PhD project, developed in collaboration with the National Gallery of Iceland, which takes the gallery’s Steina Vasulka archive as a case study. The project asks how conversational AI might support visitors’ engagement with, and interpretation of, time-based media art, and how LLMs and archival materials can be combined to open new possibilities for museum interpretation.
Steinþór Steingrímsson, Árni Magnússon Institute for Icelandic Studies
„What was playing at the movies?“ (And how OCR and small language models can help us create new research tools from old materials.)
The digitisation of printed materials on Tímarit.is has transformed access to older texts, yet the OCR’d text has rarely been usable for more than simple searches. This talk discusses recent advances in Optical Character Recognition (OCR) and how small language models can be used alongside OCR to classify and systematically extract data from historical print. The main case is the project What was playing at the movies?, which used a small Language Model to extract information on Icelandic cinema screenings, 1913–1999, from advertisements in five daily newspapers. The result – a database of nearly 40,000 films and over 300,000 screenings – reveals unexpected patterns and sheds new light on familiar history.








