'Twas the night before Christmas And all through the house Not a creature was stirring - Not even a mouse... Because everyone was watching talks from the 2025 Fantastic Futures #FF2025 conference! www.youtube.com/playlist?lis... Two days of talks on AI and machine learning in GLAMS #AI4LAM #FF2025

One of the highlights of Fantastic Futures 2025

Apply to be the 2026-7 Munby Fellow in bibliography and history of the book @theul.bsky.social www.jobs.ac.uk/job/DPZ688/m...

So sad to learn of death of Paul Ayris - a passionate, enthusiastic, encouraging and supportive advocate for #OpenAccess. The force behind the establishment of #UCLPress and the OAI Conferences, and of #OA policies at individual HEI and national funder level. RIP Paul. www.leru.org/news/in-memo...

One of the highlights of Fantastic Futures 2025

The British Library is hosting a free event on 9 Feb, 18:30: Web Archive Research in an Age of Smart Data. Hear from Emmanouil Tranos, Thom Vaughan, Dr Andrea Kocsis, and Cameron Huggett. Chair: Jane Winters. 📍Location: BL Knowledge Centre, London. events.bl.uk/events/web-a...

Want to level up your library career this year? We’re looking for international mid-career library and archive professionals to join us this summer at our International Library Leaders Programme. Learn more link.bl.uk/ILLP2026

New article: 'Seeing Library Collections Through New Lenses: The Potential of Large-Scale Digital Collections' in the Journal of Victorian Culture's Digital Forum, edited by @emilyjlm.bsky.social and Joanna Taylor doi.org/10.1093/jvcu...

Final week to submit proposals for #OpenRepos2026, don’t miss out! #OpenRepos This year, OR2026 will take place ONLINE on June 8-11, 2026. It is a great opportunity to meet the open repositories community, engage with peers and showcase your work. or2026.openrepositories.org/program/call...

Safe travels to all the #ff2025 #ai4lam visitors from all of us in the British Library Digital Research Team! See you at Fantastic Futures 2026 in Washington D.C. next September!

Closing keynote at #ff2025 from @dorothyjberry.bsky.social "What are we looking at? Massive Models and Minimal Description" starts with a quote from @ryancordell.org and discussion of luddism & AI in archives and libraries.

Loads of great talks at #ff2025 - this was fab 👇

I loved @jasonclark.bsky.social 's #FF2025 'problem statement': 'We were given access to a weirdly confident intern that can’t explain how they got their answers. It is a seamless interface that breaks search interaction + retrieval patterns.'

I love how AI projects focus on UX. It’s also worth looking at CX and how people move between systems and physical or virtual spaces, and how that connects to each organization’s culture/values. Perhaps technological developments should consider that? #ff2025

*waves* at @ukmcg.bsky.social #MuseTech25 folk from #FF2025 - it'd be lovely to see more museum presentations at #FF2026 - and the #DH2026 call for proposals is still open dh2026.adho.org/cfp/ - the conference is in Korea but remote presentations are possible! #MuseTech #AI4LAM #DigitalHumanities

Loved hearing about this AI literacy project to support information professionals (as well as the public)! misbehavingmachines.net #FF2025

This might be of interest to folk following #FF2025 #MuseTech25

If you have sample images from a museum, library or archive collection and want to see what metadata could be created about them, drop @aboutgeo.bsky.social a line! #FF2025 #MuseTech25

I'm lucky to be chairing a great session with lots of brilliant short papers, from 'The Politics of AI Training Data' (with people who critiqued the KB NL in the audience!), vision language models, work with newspapers and more! www.conftool.org/fantastic-fu... #FF2025

“Wildflowers as Benchmarks: Creating a LAM-Specific LLM Evaluation Using Historical Botanical Text” presented by @drastician.bsky.social & @miketrizna.bsky.social at #FF2025 bit.ly/3KGOs5a

Fantastic new wildflower plates and descriptions shared as a benchmark task from Smithsonian folks - github.com/Smithsonian/... with comparison of Gemini VLM models and Qwen models on it #ff2025

Slides and code from my #ff2025 talk, "Reasoning with Small Language Models (SLM) for Trustworthy Generative AI (GenAI)." Slides: docs.google.com/presentation... Code: github.com/jasonclark/a...

What would be some of the concrete and *realistic* terms/features of “responsible AI” practices in your opinion (not of the pie-in-the-sky type)? I would love to hear from folks at #FF2025 if you have thoughts on this topic. Thank you!

@[email protected] Secretary General Dr. Ines Vodopivec introduces #FF2025 short papers session bit.ly/4iOTHwf

I feel like 2025 is the year that getting structured data for catalogues from cards got real #FF2025 - so much work in orgs around the world! Come share at join.slack.com/t/ai4lam/sha...

Folks concerned about prompt optimization for tasks might want to look at packages like DSpy https://github.com/stanfordnlp/dspy #ff2025

'Distrusting, techno-pessimistic and conservative'. National Libraries and the Politics of AI Training Data by Steven Claeyssens & Sophie Ham from @kb.nl #FF2025 bit.ly/3KGOs5a

Presenting some of my work for the Mobile Subjects. Contrapuntal Modernisms today at #FF2025. In the past 3 days, I gained a lot of insight in how LAM use AI and am looking forward to apply some of their lesons to my work with @slaveryinwar.bsky.social A great way to end 2025!

In case you missed it yesterday @dchavezheras.bsky.social @kingsdh.bsky.social presented Intelligent Systems for Screen Archives ISSA collaborative project supported by BFI Innovation Challenge Fund in partnership with film archives & @kingsdigitallab.bsky.social github.com/kingsdigital... #ff2025

One more round of papers here at #ff2025 before the lunch break and poster session, where I'll be presenting our ongoing work developing a ML/AI framework for @zooniverse.bsky.social, co-authored with Hillary Burgess. For those not attending, you can see the poster here: doi.org/10.5281/zeno...

I really enjoyed being part of #FF2025 today. Very sad to miss the brilliant @dorothyjberry.bsky.social's closing talk (school performance beckons at the opp end of London), but a complete thrill to bookend the day with such a thoughtful scholar. thanks to @miaout.bsky.social for the invitation!

My semi regular refrain, #ocr is not a solved problem #ff2025 #AI4LAM

“What is the next biggest scientific discovery that can be made from a library collection (as a dataset)?” #FF2025

On a day full of conversations about big AI tech and the decisions museum, libraries and archives have to make at the intersection of licensing data for AI and their values, the AI Afterlives workbook seems like a thought-provoking exercise #FF2025 www.syntheticpasts.com/aiafterlives...

Mind blown by fashion recreation of arctic cultural clothing in 3d using PS, Illustrator, and Clo3d by this Smithsonian work #ff2025

#FF2025 in-person folk - if you're feeling a bit sniffly after travel and/or having fun in London, there are some (surgical) masks at the registration desk. I'm seeing my favourite octogenarians on Monday so I'm masking up most of the time too!

Finola Finn, Caio Mello, & Yves Maurer speaking on “Developing Archival AI chatbots: risks, benefits, and future directions” from the @bnluxembourg.bsky.social ##FF2025

#ff2025 livestream folk - if you haven't already spotted them, we have excellent transcription by expert transcribers who've seen the slides and abstracts - look for the CC option

A word cloud from the presentation transcripts would be interesting. I’m guessing "UX" "annif" & "MCP" would stand out. #ff2025

“Old system, new tricks; Using AI to Improve AV Metadata in a Legacy System” presented by Agnes Toftgård, National Library of Sweden ##FF2025 bit.ly/3MhqFJC

We began #FF2025 with two long papers - Developing Archival AI chatbots and AI to Improve AV Metadata in a Legacy System www.conftool.org/fantastic-fu... Up next, a whole bunch of different topics in an intense half hour! www.conftool.org/fantastic-fu...

Last day of Fantastic Futures 2025: AI Everywhere, All at Once begins! What an absolutely inspired community this is 🤩! Register for free access to today's livestream https://www.conftool.org/fantastic-futures-2025/index.php?page=participate and join in a full schedule of incredible talks. [1/2]

Last day of Fantastic Futures 2025: AI Everywhere, All at Once begins! What an absolutely inspired community this is 🤩! Register for free access to today's livestream www.conftool.org/fantastic-fu... and join in a full schedule of incredible talks. www.conftool.org/fantastic-fu... #ai4lam #ff2025

📣 We're proud to announce that Europeana Foundation has joined #AI4LAM. Read about this important step in our commitment to shape how AI is developed and applied across the cultural heritage sector: bit.ly/4rDfbAd #FF2025 @britishlibrary.bsky.social @bldigischol.bsky.social

for those who attended the IIIF workshop yesterday by @iiif.bsky.social this might be a useful tool to try github.com/kingsdigital... #ff2025 - developed by @kingsdigitallab.bsky.social

“AI use to identify & remove redundant, obsolete, trivial (ROT) records to reduce their digital heap.” 🙌 I think information decluttering, info hygiene, selective archiving at scale will become critical given the rising cost (IT and beyond) and energy impact of large data storage & mgmt. #FF2025

Via John Baker on the #FF2025 channel on the AI4LAM Slack: 'Cataloging books with LLMs: using out of copyright books as training data' medium.com/digirati-ch/... - following Abbey Potter and Matt McGrattan's talk

Interesting methodology for evaluating ai workflows over traditional workflows from LOC #ai4lam

Gave a talk this morning about how AI FOMO isn't a strategy and several Qs afterwards were asking if may be FOMO *should* be the driving force. But if 89% of the public think proven safety should come first, then it feels like a good idea to respect that

Probably the thing I'm most routinely asked for is information about the energy use of diff AI models. It's complicated as there's very little decent end-to-end data out there, but was very excited to see a demo of this tool this morning, developed by the French govt comparia.beta.gouv.fr/ranking