Fantastic Futures 2025: highlights and resources

Blog series Digital scholarship

Authors Rossitza Atanassova, Neil Fitzgerald and Mia Ridge

The first sign that the 2025 ‘Fantastic Futures’ conference of the AI4LAM Community was going to be special was when our AI Everywhere, All At Once’-themed Call for Proposals received hundreds of submissions. And then tickets for in-person attendance sold out within two and a half days... 

By the time the conference started in early December, we had over 600 registered participants – more than half of those online – showing just how much the AI4LAM (AI for libraries, archives and museums) community has grown over time as interest in AI and machine learning increased. 

The 250 international participants onsite at the British Library Knowledge Centre, 3-5 December 2025 included dozens of presenters who delivered 16 workshops, 19 posters and 45 talks over the three days. The talks on Thursday and Friday were also livestreamed to hundreds of online participants.

Find and share the following conference resources:

• All the abstracts, slides and posters presented at the conference
• Livestream recordings playlist AI4LAM YouTube Channel
• Official photos on Flickr

What did we talk about at FF2025?

In his welcome message Jeremy Silver, British Library interim CEO, congratulated the community on the conference name: ‘Fantastic Futures’ highlights our collective collaborations and networks enabling exciting AI research and sustainable practice for the greater good.

Rachel Coldicutt's opening keynote called for responsible and ethical uses of tech that ‘works for 8 billion people, not 8 billionaires’. She shared the Careful Consequence Check tool, a guide ‘to make a rapid assessment of the potential risks and issues involved in using or adopting an AI-powered product, feature, or service’. Her talk provoked discussion about the pressure on GLAM institutions to innovate and keep up with technological change, and the crucial role of domain expertise in delivering responsible, ethical and values-first AI.

Topics discussed over the three days ranged from the huge challenges – environmental and economic sustainability, ensuring alignment with our missions, ethical and responsible AI, human-centred AI, ensuring value for money – to practical approaches – evaluation, scalability, cyber security, multimodal collections – and throughout it all, managing the pace of change and impact on staff.

Our closing keynote speaker Dorothy Berry powerfully centred people in the archive, including the importance of professional skills and values, domain and subject expertise coupled with the huge responsibility that comes with using AI. She highlighted some of the pitfalls and compromises in AI-assisted solutions for archival description and online discovery, the complexities of working with large historical archives in the context of multi-institutional digital initiatives, and the tensions between human and machine-generated knowledge. Her description of the roles that GLAM (gallery, library, archive and museum) staff inhabit in digital projects resonated with many.

Any attempt at summarising FF2025 will be incomplete but some highlights and themes include:

• Insights from partnerships with AI companies (Bodleian and OpenAI), and the many challenges of introducing AI into workplaces: ‘what staff want their CEO to know about AI, and what they're not telling their CEO!’
• Groundbreaking tools like comparia.beta.gouv.fr with carefully designed user interfaces to encourage public awareness of LLMs uses and limitations
• Operational aspects of responsible AI, such as the hardware resources discussed by Christoph Poley, National Library of Germany 
• Copyright and AI is still a lively topic with some innovative approaches (CORAL), tied into other questions of licensing data, the changing role of open data and the many impacts of bots scraping GLAM websites
• Evaluations of the promising use of specific AI models and workflows for metadata creation, image similarity search, discovery and creative reuse, and the role of staff in pinpointing bottlenecks where AI could deliver the most value.
• Work that places users at the heart of GLAM work, including user surveys and UX e.g. OED AI Assistant, AI in the Library, and the AI Afterlives workbook
• Detailed overviews of production pipelines using AI (e.g. RAG) in combination with other standards and technologies, such as Multispectral Imaging, IIIF, automatic text recognition, BIBFRAME and Linked Open Data
• Rasa Bocyte’s lightning talk about the Europeana/Beeld & Geluid Culture for AI study that defined the uncertainty and ambiguity that the GLAM community feel towards AI as our 'superpower'.
• Learning why Hawaiian shirts can make it difficult to detect chyron (those running lines of text) in AAPB broadcast news.
• Finally, the amazing audience participation during Q&As and poster sessions!

As organisers, we wanted the conference to introduce and invite participants to join the AI4LAM community at future events and on the AI4LAM Slack. So for the first time at a Fantastic Futures event, we included interactive poster sessions designed to encourage conversation, and we organised ‘Dine Arounds’, informal dinners to help attendees meet each other in small groups. (We borrowed this idea from the newcomer dinners ACH used to organise at the international Digital Humanities conference and it worked well!)

We hope that all participants learnt something new, developed an idea, had some of their questions answered or simply enjoyed conversing with old and new friends.

With thanks to everyone who put in a proposal (whether it was successful or not), the presenters of workshops, talks and posters, the livestream and transcription teams, our expert proposal reviewers, and all the BL staff (especially the International Office, Digital Research, Finance and Events teams) who helped make FF2025 such a success!