The current anxieties around the appearance and embedding of AI and LLMs in the University sector and GLAMA (Galleries, Libraries, Archive, and Academia) sector triggered a strand of research in University Cultural Collections at the University of Leeds. This included two AI focused short fellowships in conjunction with the Library and Digital Creativity and Cultures Hub (DCCH) in 2025 entitled. University Cultural Collections and Generative AI: Critical Approaches in a Library Context . Their report called ‘for the informed use of Generative AI by cultural practitioners going forward, which could be fostered through institutional support, standardised training, and open-source tools’. Building on this recommendation, we began to model the role library and archival collections could play as the basis of future knowledge and discovery, through the potential use and origination of AI tools. This approach is grounded in a series of critical questions around how AI technologies are used to inform researchers of archival collections’ content and value, how and where technologies are valuable, and underlying concerns around sustainability, professional values, ethics and inclusivity.
To further explore and iterate on our previous work we launched our current DAIMs project which concludes in July 2026. The project was designed to develop an ethically informed framework for the use of Generative AI and LLMs in relation to our Cultural Collections. The project aligns with the UoL Digital Libraries Infrastructure Project (DLIP), AI governance, Open Knowledge and our commitment to equity of opportunity for researchers. It responds to growing sectoral anxieties around the deployment of AI tools and LLM approaches in relation to collections and the current lack of strategic guidance for researchers and students necessary to underpin a healthy research environment. It specifically focusses on the potential use of tools (those currently approved by the University), the prototyping of new open-source tools such as the metadata agents, and methods for the creation and tracking of metadata to operationalise library processes for cultural collections and to enhance discoverability and linkage. It is informed by a series of collections-based experiments and built around a diverse and interdisciplinary team including collections staff, digital humanities scholars, researchers and our RSE community.
This paper will unpack the findings of this very recent project and consider what is next in terms of developing common, shareable resources and approaches. It will also evaluate the collaborative nature of the working processes, considering the value of building interdisciplinary teams to consider foundational questions and the role DH methods and approaches play in this context. We will present the results of our experimentation and share our emergent framework.