Session 2

Session 2: Unlocking the AI potential

Wednesday 6th July – 14.45-16.15

Chair: Anna Clements, University of Sheffield, United Kingdom

2.1 Leveraging artificial intelligence to explore and analyse digital collections: a roadmap for the national library of France, Emmanuelle Bermes, Bibliothèque nationale de France, France

Artificial intelligence has become a mature technology, widely used by private platforms in the cultural sector such as Amazon or Netflix, making its way in our day-to-day lives through social networks and virtual assistants, and also taking up as a tool for the digital humanities. However, its use in libraries is still at an experimental stage. In order to improve their skills and refine use cases, libraries such as the BnF need to partner with academics through R&D projects, which leads to proof of concepts and scientific analysis of various digital corpora. Then, it becomes possible to envision an industrialization of AI with the goal to improve library services towards their patrons, in particular in the research field. 
Building on the projects we’ve been conducting the past few years, always implying a close collaboration between librarians and researchers, in 2021 the BnF framed a roadmap for the development of artificial intelligence. Our new DataLab, opened in October 2021, is an asset in growing our experience with this technology. But complementary tasks need to be undertaken, including a closer look at ethics, improvement of librarians’ skills, development of a suited data infrastructure… Artificial intelligence has the potential to transform activities in every area of the library, from cataloguing to public services, from collections processing to data analysis, from building conservation strategies to generating enriched data based on the digital collections. The BnF roadmap for AI includes a programme consisting of several projects, most of which will have a strong impact on how researchers interact with our digital collections. 
This presentation will provide an overview of the BnF roadmap for AI, with a specific focus on those areas that are fostering collaboration with researchers willing to explore our digital collections: computer vision, HTR (handwriting text recognition), metadata generation or enrichment for massive digital collections such as web archives, and finally personalized recommendation of cultural content, the latter being designed with an ethical approach. We will discuss the actions that the BnF is planning to take in order to enact the transformation of the organisation, including professional training and collaborations with partners in the academic landscape. Finally, we will present our first concrete results with AI applied to heritage collections. 

2.2 Research project leadership: perspectives from a cross-organisational case

Developing capacity for leading research projects at the university library is of utmost importance for meeting current and future demands to the library as a sparrings partner with the university, responding to rapid change in research environments and ultimately maintaining the relevance of the library. Libraries need research leadership capabilities as they move away from their traditional principal function as collection builders and providers of content to a more strategic community-oriented and collaborative role. 
In this paper we explore the perceptions and requirements we met in driving a cross-organisational research project and consequently the opportunities and challenges we identified for research leadership at the library. 
The case is a project in which we are investigating the extent AI-powered search systems support scientific and research integrity in literature-based research methodologies. The investigation itself was a sequential test design consisting of three stages. These were: 1) desk-top research to identify relevant AI-powered search software, 2) think-aloud tests of the functionality of a selection of AI-powered search software and 3) a search-hackathon to assess the quality of search output. Our findings may or may not result in the purchase of an AI-system and development of new services at the library. 
This project exemplifies a new approach for library management, where laissez-faire leadership style allowed the project group to use their creativity, resources and experience to design and run the project and also make mistakes. Thus, time was invested in the project and trust invested in the project group’s abilities. The outcome includes a transparent and reproducible approach to product testing, publication in scientific journals as a predefined endpoint and likewise acceptance that the result of this years long project may result in a negative outcome. 
Implications point to learning opportunities, multi-skilled teams and leader development, meaning that understanding the scientific process where a longer perspective and freedom of research in projects based at the library will result in future-focused services, robust results and ultimately strategically developed partnerships with research communities at the university. 

2.3 Engaging in research and cultural heritage collaborations to endorse AI and machine learning activities, Liisa Maria Näpärä, Päivi Maria Pihlaja, National library of Finland, Finland

This paper discusses the role that the research and project services could play in outlining the position of a GLAM actor like national library towards artificial intelligence (AI) related initiatives and service development. The paper focuses on recent activities at the National Library of Finland (NLF). We describe the variety of activities within the library, ranging from infrastructure services to research related needs, and also consider their embeddedness in wide-ranging partnership networks. A central question is how unshared understanding regarding AI within the NLF influence its role in this development. 
The National Library of Finland (NLF) has become increasingly engaged in various research projects in and between different disciplines. During the last couple of years, the library has taken coordinated measures to developing its practices related to these. Recently, researchers’ experiences from collaboration with the library were systematically collected. Needs of the researchers related to the digital collections of the library were mapped, and our practices were benchmarked against other national libraries. 
We will firstly present a number of data-driven collaboration projects that have taken place at the library during 2021 with a digital humanities approach. While others have been more focused on analogue collections, some of these have developed new tools for data processing or otherwise promoted the use of artificial intelligence. New knowledge and experiences gained from these collaborations permitted us to deepen our understanding of how to foster such activities internally. 
Now focusing on AI and machine learning, we performed a qualitative analysis relying on internal information collection at the NLF gained by means of several expert interviews. These exchanges provided us with an interesting overview of the large networks consisting of both national and international partners or other external stakeholders in which the NLF operates. Examples of these activities reach beyond the library sector and research activities. They concern, to mention a few examples, infrastructures facilitating open access to publications and cultural heritage materials, or dataset services and the long-term preservation and findability of these by sustaining and improving the quality of metadata and identifiers. New opportunities related to the use and promotion of the machine learning methods and interoperability of systems have also unfolded through partnership agreements and targeted collaborative research efforts. 
In this paper, we will describe our findings. We will present our analysis the variety of internal activities along with the large cooperation networks and joint interests that already exist between GLAM organisations and with researchers on national or international levels. We maintain that the future role of AI at the NLF could only be apprehended as part of larger partnerships and collaborations. This also applies to development projects within the library. We seek to delineate measures that may be helpful in modelling internally a shared outlook towards these broad and complex issues. 

Lighting talk:

Academic Libraries as Hubs for AI Based Culture Diplomacy, Adam Sofronijevic, Aleksandar Jerkov, Dragana Jankovic, University library ”Svetozar Markovic”, Serbia

Lightning talk introduces the idea that academic libraries are ideally position to harness the power of AI and use their established reputation and wide access to researchers, students and citizens in general to provide for meaningful culture diplomacy activities based on AI. This idea is presented based on University library in Belgrade experiences and activities in this area as well as on considerations based on insights coming from literature on libraries, AI, diplomacy and other relevant and related fields of social and technological interactions among peoples, institutions and individuals. Culture diplomacy activities around the world are not using explicitly the power of AI as of first half of 2022 in spite of AI being a potent indicator of all things allowing for global soft power projection. Meaningful culture diplomacy provide not just for one sided influence such as soft power projection, but for productive dialogue among two or multiple parties. AI based culture diplomacy activities are ideal for this kind of dialogue since AI technologies allow for individual approach to citizens in global multicultural environments. Academic libraries have the potential to become hubs of such activities because of their unique and equidistant position between science, technology, culture and citizens. The lightning talk introduces the possibilities for academic libraries in this area, explains details of activities conducted by University library in Belgrade that are considered illustrative and building on this basis propose some scenarios for globally productive approach adding value for all sides involved in the process of culture diplomacy activities based on AI that might be originating from academic library hubs. By building technological and organisational basis for conducting meaningful activities based on AI across multiple countries and regions academic libraries become not just research oriented institutions that discover new possibilities and relations among entities from different societies and cultures, but also provide grounds for engagement of citizens and international collaboration among individuals and institutions.In discovering new ways to connect AI technology owners and developers, data providers, data creators and users of machine learning applications and tools academic libraries might not just reinvent themselves in the new environment of advanced phases of fourth industrial revolution, but also find new meaning in not just serving one community, but truly serving global users across several countries and regions and between cultures, religions and languages.The possibilities for various kinds of influence of AI technologies and applications on culture activities are often highlighted, but the concrete implementation especially in building relations among different nations and people is still lacking. Academic libraries might be the right places to bridge this gap, but they need to become true hubs merging AI technology, citizens and organisations across several countries or regions in the meaningful actions that add to productive dialogue and increase in understanding and communication between parties that otherwise might not be making such positive steps forward so easily. 


52nd LIBER Annual Conference