I have been engaging with Professor Mark Humphries, who writes the excellent Generative History substack. Mark has analysed the capabilities of Gemini 3 for HTR. In this Bluesky post I add to his findings. @timhitchcock.bsky.social #history #ai #llms generativehistory.substack.com/p/gemini-3-s...

Gemini 3 Pro in Consumer Bot mode can identify and markup text regions, identify lives, and label lines by region. Moreover it can pretty accurately transcribe English language C17th legal depositions, despite a complex layout. Follow link for the prompts I used gemini.google.com/share/c42566...

I am relaunching the ai-and-history collaboratory I ran last December. First meeting Tuesday, December 9th 2025. Then roughly monthly by ZOOM. Let me know if you are interested. github.com/Addaci/ai-an... #ai #history @timhitchcock.bsky.social

I am looking for a couple of beta testers who will take half an hour and try out a Socratic Research Assistant I have been building powered by Gemini 3.0 Pro Preview. The idea is to get much greater control over the persona, and inference approach of Gemini through the interface

Here's my agenda for the next three months: generative-ai, historical research and archival practice: generativelives.substack.com/p/whats-top-... #earlymodernhistory #history #archives #ai #ml

I am starting a Substack called Generative Lives. This offers a user's perspective on generative-ai and its potential application to historical research and archival practice. The Substack is free and seeks to encourage hands-on experimentation. open.substack.com/pub/generati...

Hello BlueSky. I haven't been active for a while. I am looking for examples of museums, libraries and archives doing interesting things with generative ai and would welcome suggestions of websites to look at. #earlymodern #archives #libraryscience

Did ships get bigger in the C17th? Yes. Contact me if you would like access to the data and would like to discuss. And of course the devil is in the detail, but this is a relatively and well defined large dataset #earlymodern #marinehistory @thiagokrause.bsky.social @brodiewaddell.bsky.social

A potential use of AI image analysis which could be done at scale as part of an analysis of historic literacy? I loaded images of 140 signature, initial(s), mark(s) into NotebookLM & interrogated them palaeographically (with zero training). @thiagokrause.bsky.social @timhitchcock.bsky.social

Looking for a Turkish speaker who would be willing to sample a couple of minutes of a Turkish podcast generated by Google NoteBook LM Plus [podcasts in 50 languages, handles 300 sources, very useful mindmaps]. Can send .wav file or give NB access @thiagokrause.bsky.social @timhitchcock.bsky.social

Looking for two or three Early Modern marine historians, marine archaeologists, and knowledgeable enthusiasts interested in exploring novel, as yet unpublished 5,000+ ship database [SOLM-QUANT] for C17th commercial (and naval) ships - Get in touch to learn more #marinehistory #earlymodern

I have been busy these last four months. Now looking for a publisher. Any suggestions. Can supply draft chapters and appendices for anyone interested in taking a look. @timhitchcock.bsky.social @brodiewaddell.bsky.social

Enrollment is now open for the 4th edition of the Master's in Computational Social Sciences at UC3M! No prior knowledge of programming or stats is needed—perfect for social sciences grads. Questions? Contact us or ask to be connected with our student ambassadors! Please, share!

We have a nice new shiny organization card for MarineLives on Hugging Face, where you can find models and datasets huggingface.co/MarineLives

I am working with @beismo.bsky.social to fine-tune two small large language models to correct raw machine transcription of Yiddish in Hebrew script. We plan to show the results this coming Tuesday at our next ai-and-history-collaboratory @timhitchcock.bsky.social @mark-l-thompson.bsky.social

Do you wonder what "fine-tuning" a large language model means? How could it help in historical research? Next Tuesday during our weekly one hour ai-and-history-collaboratory we are going to run some code and train a small large language model called mT5 github.com/Addaci/marin...

Great to see our ai-and-history-collaboratory is having some impact. And thanks to @mark-l-thompson.bsky.social for his participation. The topic for next Tuesday is fine-tuning large language models for use in historical research github.com/Addaci/marin...

Interested in learning to use large language models for historical research? We have a Colab notebook which guides non-coders through using APIs to interrogate data github.com/Addaci/Aurel...

We have a Google colab notebook as background to session two of our ai and history collaboratory. If you are interested please try some of the exericises - no code required github.com/Addaci/marin...

We are holding the second session of our ai-and-history collaboratory on Tuesday, December 3rd 2024 at 4 pm London time. We will be learning how to use APIs to control large language models for historical research. github.com/Addaci/marin...

Curious about our ai-and-history-collaboration? We hold one hour ZOOM dropin sessions each Thursday at 4 pm London time. Direct mail us if you are interested in finding out more. github.com/Addaci/marin...

1/ The first session of the eight session ai-and-history-collaboratory took place yesterday. Great turnout and stimulating discussion of how to prompt large language models in the context of historical research. github.com/Addaci/marin...

Our ai-and-history-collaboratory kicks off this Tuesday for the first of eight sessions. This Tuesday we are discussing Prompt Engineering. github.com/Addaci/marin... #skystorians #machinelearning

Great article on training librarians in digital humanities

An exceptionally helpful post by @burtenshaw.bsky.social I have changed session 4 of our ai-and-history collaboratory to tackle data broadly, including creation and use of synthetic data github.com/Addaci/marin...

The slides and video from my talk yesterday at the IHR Digital Seminar on ai and history ihrdighist.blogs.sas.ac.uk/2024/09/tues...

I wrote the book I wish I’d had when I was starting out. It’ll be available as paperback or ebook next month. Sharing the Amazon link here because it includes the reviews / testimonials from the back cover. #history #AcademicWriting 🗃️ www.amazon.com/Craft-Histor...

Fascinating primary source on Dutch/Venetian relations in early C17th. I am going to read this carefully and cross-reference against our growing High Court of Admiralty Corpus from the 1610s and 1620s #skystorians #earlymodern

I am compiling a list of potential sources of English language history related training data & metadata to fine-tune large language history domain models. They have varying licencing terms. What am I missing? @danielvanstrien.bsky.social github.com/Addaci/marin...

1/ Six more slides from my talk next Tuesday - the order may change, but let me talk you through them

This has to be the most fun post I have seen today. If I were a seagull I would definitely head south!

I am mulling over the messages I want to convey when I talk this Tuesday about ai+history. I'm not sure I am going to make friends (or I am right) but here goes @tedunderwood.me @michielree.bsky.social @javiercha.bsky.social @timhitchcock.bsky.social @bschmidt.bsky.social @zoeleblanc.bsky.social

1/ The first six slides from my talk this coming Tuesday on ai and history - a new set of skills for historians ihrdighist.blogs.sas.ac.uk/2024/09/tues...; see also our collaboratory which I am launching at the talk: github.com/Addaci/marin...

Looks like a great course and a grat opportunity to work with @cathrynpearce.bsky.social an early @marinelives.bsky.social volunteer and supporter

A treasure trove

When our Print & Probability collaborator UCSD grad student Nikolai Vogler heard that ESTC was down due to the BL cyberattack, he jumped into action. Here’s the site he built to help #earlymodern scholars access the essential metadata needed for teaching and research 📚 📜 estc.printprobability.org

My Latin master Phil Hiscock (1969) would have loved Google Translate. If you would like to prep up on a more modern tool for the humanities (which Phil would have loved) join our ai-and-history-collaboratory: github.com/Addaci/marin...

And a huge round of applause to @dhidalgo.bsky.social my 1000th follower. Beware, you are going to hear a lot about the C17th English Admiralty Court and Ai-and-History. But I promise you won't be bored!

Please help me build this list go.bsky.app/6RfFHkd @thomwolf.bsky.social @tedunderwood.me @timhitchcock.bsky.social @michielree.bsky.social @danielvanstrien.bsky.social @jwbaker.bsky.social @adamcrymble.bsky.social @dorialexander.bsky.social @transkribus.bsky.social

Inspired by a couple of conversations on @bsky.app over the last two days, we have changed the goal of our ai-and-history-collaboratory to be more inclusive, opening it up to all students and public historians, especially zero coders, with open arms github.com/Addaci/marin...

How about it? @idlemrb.bsky.social @kimberlybiscuits.bsky.social @kejamieson.bsky.social @michielree.bsky.social @mark-l-thompson.bsky.social @baliscoe.bsky.social @danielvanstrien.bsky.social

I am in blatant recruitment mode. We have 7 Blueskyers signed up for our ai-and-history-collaboratory. We have 3 slots left. Forget the "we are targeting graduate and doctoral students". Care about history. Inquisitive about tech? Want to engage. DM us and your are in. github.com/Addaci/marin...

We are currently working on issue 43: Women in Ports! This issues' guest-editor is Irene Jacobs from the Martiem Museum Rotterdam.

Another slide from my talk next Tuesday on ai assistants and agents: a new set of skills for historians. A prescient article (2013) by @timhitchcock.bsky.social and a facinating book by @adamcrymble.bsky.social ihrdighist.blogs.sas.ac.uk/2024/09/tues...

This should get the discussion going at my talk next Tuesday on ai assistants and agents: a new skill set for historians? ihrdighist.blogs.sas.ac.uk/2024/09/tues... #skystorians #machinelearning

Not sure why suddenly I have an "invalid handle". This has been my handle on Twitter, and now Bluesky for 12 years. Reddit fix doesn't seem to work. Can you guys still see me?

What a fabulous article. Thank you for reposting this @ideesdupasse.bsky.social. When I workd in teh pharma sector I took a senior R&D group from Glaxo to Shanghai and Beijing in 1997 and was deeply impressed by the hisorical botanical, pharmacological and medical practices