Parallel Session 1 – Charting the Field of Artificial Intelligence for Research Libraries I
Moderator: Birgit Schmidt, Göttingen State and University Library, Germany
Location: Tassos Papadopoulos – Room 101 (1st Floor)
1.1) AI Augmented Library – Leading Institutional Generative Artificial Intelligence Delivery
Presenter: Masud Khokhar, University of Leeds, United Kingdom
Since ChatGPT has taken over the world by storm in late 2022, Generative AI (GenAI) has been a topic of high intrigue, excitement, and anxiety in higher education institutions. The human-like content generation capabilities and the conversational interfaces of GenAI tools have invited philosophical and practical debates on the ethics, biases, practices, and productivity promises across research, education, and civic engagement missions of the Universities. GenAI challenges the current norms of learning and assessment design in institutions, along with potentially creating major opportunities for workflow and task optimisations. It also creates an opportunity for libraries to position themselves as leaders in GenAI delivery.
Adoption and governance of GenAI in higher education institutions has been a difficult task. The fast-changing landscape means that institutional governance and regulation processes can’t keep up with the pace, and adoption requires significant levels of staff upskilling. In addition, the simple to use interface and the cross-cutting implications of GenAI open new opportunities for owning digital transformation across the institution, moving beyond the traditional IT boundaries. The role of libraries here is both fundamental and transformational, particularly in relation to a holistic approach towards GenAI, considering the challenges, biases, EDI issues, and impact that the technological advances bring with them.
This presentation will provide a case study of how University of Leeds Libraries have positioned themselves as Library AI leaders in the UK through a series of steps. The presentation will initially focus on the outcomes of a research report, which was commissioned by the library called “Looking towards a brighter future: the potentiality of AI and digital transformations to library spaces” and the impact it has had so far in how the library operates. It will then highlight how the library was crucial in the formation of two institutional working groups on AI in research and education, namely Working Group on AI Technology in Research (WAITR) and Working Group on AI in Student Education (WAISE). The working groups have already delivered key outcomes, in particular the policy for use of GenAI for staff, co-production of GenAI student guidance, and the conception of the first GenAI summit at Leeds. The presentation will also highlight some key lessons learnt in this area, in particular relating to the change of pace in GenAI developments against the governance arrangements for GenAI delivery and the approach towards AI literacies and AI experimentation. It will also focus on how the library has positioned itself as a major entity in the GenAI governance, bringing EDI, ethics, and user experience as key focus areas of discussion. The presentation will provide a roadmap for GenAI delivery in the library, and how it has led to a rethink of the structures through which we deliver digital services and initiatives. The presentation will also cover how we are partnering with other leading and like-minded libraries to harness the full potential of GenAI, for now and in the future.
1.2) Getting Academic Library staff up-to-speed on Artificial Intelligence
Presenters: Beate Granström, Chalmers University Library, Sweden and Helena Schmidt Burg, Gothenburg University Library, Sweden
With the public release of generative artificial intelligence (AI) tools, most notably ChatGPT in 2022, interest in the technologies has swept across various organizations like wildfire. Higher education institutions and their libraries are no exception. Many academic library staff have found ways to develop their competence around AI and investigate what AI development will mean for academic libraries. Yet, if developing a better understanding for (generative) AI at academic libraries broadly is a goal for the organization, then the question arises of how a large proportion of a library’s staff can be reached, rather than a few engaged individuals, and what aspects of AI will then be relevant to address.
These are questions that we have faced in a collaboration between three libraries at universities in western Sweden. In the presentation we will share experiences from this collaborative effort to further professional development among staff members concerning AI in libraries. The program runs during the academic year 2023/2024 and contains several online lectures and two active learning classroom (ALC) sessions. The lectures were streamed live to all staff and were also recorded and made available for the three institutions. The ALC sessions, which allow for active engagement with the technologies as well as other participants on campus, required registration and were attended by a subset of the staff.
Although not all staff attended all or even parts of the program, it aimed to reach library staff beyond those who are particularly motivated to learn about AI. Apart from facilitating participation by making it less reliant on participation and time, the lectures and ALC sessions connected closely to issues of relevance in academic libraries, rather than at the university at large, and in different areas of responsibility to boot. They also linked to the local situation at the three universities, thus making them more immediately useful for the everyday work tasks. The fact that the sessions were planned and, for the most part, implemented internally has facilitated maintaining a local focus. Aspects of AI have been addressed from the perspective of their relevance to various library stakeholders (students, researchers, library staff), ongoing development projects have been presented, and issues around bias, privacy, security, and AI literacy have been discussed.
In the presentation, we will discuss the challenges of designing content for professional development around AI at a time when so much is happening and happening fast, at the same time as much actual implementation in the library field is in its infancy. We will also share our approach to these challenges, with a particular focus on how to make the content relevant to academic library staff with varying professional backgrounds and working within different specialized areas of responsibility. The presentation aims to raise awareness of, but also problems involved with, how professional development can be conducted within an area such as AI which is at the same time ubiquitous and can be viewed as a very specialized area of expertise.
1.3) Bridging the Semantic Gap: Innovations in AI-driven Access to Czech Digital Libraries
Presenters: Filip Jebavý, Moravian Library in Brno, Czech Republic and Filip Kersch, Library of Czech Academy of Sciences, Czech Republic
In recent years, the digitization of documents within Czech libraries and archives has reached unprecedented levels, presenting new challenges and opportunities for information retrieval. The traditional full-text search methods employed in the Kramerius digital library system, while effective in locating words or phrases, fall short in finding images or capturing the semantic nuances and meanings within the text. This proposal outlines two groundbreaking projects, Orbis Pictus and Sémant, that aim to revolutionize the accessibility and exploration of digitized content within Czech libraries, transcending the limitations of conventional search methods.
SemANT Project: Enhancing Meaningful Search and Navigation
The SemANT project is dedicated to overcoming the limitations of conventional full-text search methods by introducing a semantically enhanced search system. This approach allows users to delve beyond specific words and search for meaning and context within the text. Through the incorporation of automatic topic identification, users can seamlessly navigate related documents and explore areas of interest. The project also allows users to search by text segments (e.g., paragraphs) and to specify their own topics, creating a personalized and intuitive search experience.
Orbis Pictus Project: Unlocking Graphic Content through Machine Learning
While textual information dominates library collections, a significant portion of cultural heritage lies within graphic elements, including drawings, maps, photographs, and diagrams. The Orbis Pictus project leverages machine learning methods to identify, categorize, and contextualize these graphic elements within digitized documents. By extending the capabilities of digital libraries to include graphic content, the project enhances user access and opens new possibilities for creative industries. By bridging the semantic gap between textual and graphic content, Orbis Pictus contributes to a more holistic and inclusive exploration of our cultural heritage.
In this lecture, we will delve into the methodologies, challenges, and anticipated outcomes of these projects, showcasing their potential to redefine how users interact with digitized materials within digital libraries. The innovations presented in Sémant and Orbis Pictus mark a significant step forward in leveraging AI for a more enriched and meaningful exploration of cultural and historical artifacts. Additionally, the presentation will explore the necessary changes to the core of the Kramerius digital library system and its user interface. We will discuss how these enhancements align with the broader goal of transforming the user experience and expanding the capabilities of digital libraries in the context of evolving technologies and user expectations.