Approaching AI at the National Library of Scotland
CILIP: The library and information association
We represent and champion all information professionals
Robert Cawston, Director of Digital and Service Transformation, introduces a new AI Statement for the National Library of Scotland.
This article is featured in the AI hub
Approaching the hype cycle
Despite the rapid developments in large language models (LLMs) and generative AI tools over the last 12-18 months, for many we remain at the “jagged frontier”(1) of AI capabilities.
LLMs can be excellent at some tasks that seem hard to humans, and bad at some tasks that seem easy – and it is difficult to know the full extent of capabilities or when they might fail.
Commentators are noting a levelling off in the rate of improvement with models unable to sense check own work and struggling at handling(2) outliers.
Companies have piloted new tools but are not moving beyond an early stage(3) and users are yet to embed AI tools into everyday habits, often saying, "very cool, but not for me".(3)
Amidst the hype cycle of new technology and utopian/dystopian predictions it can be hard for libraries to know where, when and how to test and adopt new technology.
Many continue to approach AI with caution and, as the technology strategist Rachel Coldicutt noted in her keynote at this year’s Scottish AI Summit, “FOMO is not a strategy”.(4)
Careful thinking
As a national library we have a role to play as a trustworthy source of information and a responsible steward of collections and data. This requires us to think carefully about how and when new technology is used.
We already have a strong track record of experimenting with AI tools through collaborative research(5) and engaging in wider debates around new developments.
Our AI and Machine Learning Symposium in April 2023(6) brought together expert speakers from libraries, universities and the creative industries, and the Data Foundry(7) enables collaborative projects around our open data collections.
Our AI statement was created as a guide to the Llbrary’s decision-making on how we employ AI tools, communicate about our use of AI, collect AI-generated material, and allow our data to be used by others.
Responsible adoption
The statement’s emphasis on “responsible adoption” aims to strike a balance between the risks and opportunities presented by artificial intelligence.
We recognise the possible harms presented by many AI tools – from data protection issues and copyright infringement to digital exclusion, patterns of bias, and environmental impacts.
We also want to remain responsive to how technologies could improve how we work, including opening up access to content, automating processes around the management of collections and enabling new forms of research and creativity.
领英推荐
This approach, set out across eight commitments in the statement, is perhaps best summed up by the aim to foster a culture of “curiosity and care” across the Library – a phrase gratefully borrowed from John Fitzgerald at SCVO.(8)
Looking ahead
The AI statement was co-created by a group of Library staff, including representatives from metadata, development, digitisation, collections and research, maps, rights, and digital scholarship.
This group continues to share updates, ask questions and identify potential areas of impact across the organisation.
We’re hoping to open up the group to new members, including external experts and Library users, to regularly review the statement and identify new research and partnership opportunities to pilot AI tools and methods.
For further information and any questions, please contact: [email protected]
Further reading, references and links
(3) The AI summer
Keep in touch
Sign up to the CILIP newsletter to get the latest updates on how CILIP is advocating for libraries, and the latest news impacting our sector.
Librarian
3 个月Very nice
Retired Cataloguer and Museum Volunteer
3 个月The practical application of the Cataloguing Code of Ethics? https://tinyurl.com/3v567ec5 I think as a profession we all recognise the common issues facing us and our communities. Good to see most of these embedded in the Statement ?? Access to resources and metadata Acknowledging bias Advocacy Collaboration Critically applying standards Diversity, equity, and inclusion Education and training Respect for agent privacy and preferences Responsibility and transparency Understanding and meeting user needs