David Smith

While in Paris, I’m giving a talk at Inria on Nov 6 (also on Zoom!) about my PhD work on how humans and machines understand concepts, from mental imagery in text to how "equivalent" words differ in meaning across languages and cultures. Abstract & details in the image 👇

I’m at #EMNLP2026 until Oct 30, then in Paris until Nov 8! Old friends and new: if you’re in either city, happy to chat about research or just grab coffee/food/drinks/dessert! I’m also on the lookout for postdoc and industry positions, so feel free to reach out about that too!

JOB(S): My dept has an open rank open specialties (w/in Information Science / Library & Archival Science & closely related) search out of which we hope to make multiple hires! Im not on the committee but can say we'll have new teaching needs in social impacts of information technology & also ethics!

Here at @schmidtsciences.bsky.social, we now have full abstracts up on our website for the 19 new HAVI Awards! These are interdisciplinary DH projects that are improving AI as a tool for new kinds of humanities research. Congrats to all the awardees! www.schmidtsciences.org/havi-2026-an...

Find us at #EMNLP2026 to learn about narrative qualities of pretraining data! 🌟 📊 Via our Explorer tool, you can interact with 13M pretraining passages and 11 narrative dimensions, plus upload your own data for analysis: cls-lab.colorado.edu/tejo9855/dol... We'd love to see what you discover! 👀

Really nice analysis by @wouterhaverals.bsky.social et al. of a “schoolbook canon” of prosody using text reuse and our passim tools!

Oulipan deformance avant la lettre

I’m really happy to see @si-wu.bsky.social ‘s paper out now. Taking advantage of literary works, where we often have multiple very different translations, for both training and evaluation, is very exciting!

I will be in Budapest for EMNLP and WMT. Happy to chat! I am also on the job market for postdoc and industry positions. 📄 Paper (more findings on semantic similarity and filtering): www2.statmt.org/wmt26/pdf/20... @dasmiq.bsky.social

Literary works are often translated into the same language multiple times, and each translation interprets the original in its own way. How can we use these different translations to train better literary MT systems? Can synthetic references be as good as human ones?

apply.interfolio.com/195460 new post-doc at Princeton: AI as a "cultural technology" — its role across arts & humanities & knowledge infrastructures, AI's societal effects on creative & cultural industries (literature, music, film, visual arts). Looking humanistic approaches to AI systems!

There is a very powerful narrative in the history of the humanities about the alleged centrality of bildung rather than research that only works if you ignore, as that narrative does, the profound imbrication of the practices of reading and writing that came to be 'humanistic' in government.

I wrote a whole book about this! Research is and has always been central to the humanities! It's not a late breaking addition that emerged in response to the research culture of the natural sciences. yalebooks.yale.edu/book/9780300...

Too many people are calling on the community to "change how we value papers". Not enough people are posting "this is what we look for in our next hiring round".

“AnaChartiste”— to appreciate this, you need to know that the trained archivists who graduate from the École des Chartes have a reputation for being rigid, rule following etc. So that they* have been angered to anarchism… is something

This talk by @himself.bsky.social was wonderful. Models mechanize, map, + remix the "house of intellect", connecting together previously disconnected hallways of knowledge. LLMs are not stochastic parrots but mechanical bards. Cites: Jacques Barzun, Albert B. Lord, Nestor Guillen, Herbert Simon

Cool job alert, this looks excellent: Cornell is hiring up to 5 postdocs in humanities/social sciences & AI. 2 year positions. jobs.chronicle.com/job/38045266...

On the 20th anniversary of Zotero's 1.0 release on October 5, 2006, I wrote a history of the collaboration and years it took to create the popular research tool now used by over 20 million people: newsletter.dancohen.org/archive/the-...

Postdoc (18 months) at Inria Paris! Let's go back together to the manuscripts behind Dees' Atlas of Old French, with CATMuS & CoMMA, and measure how editors shaped our dialect maps. Skills: Old French, palaeography, TEI. Latin a +. Deadline 30/11 recrutement.inria.fr/public/class... #digimed #DH

We've been super busy the past few months getting out some research on the computational analysis of film, so I've put together a page with the highlights. people.ischool.berkeley.edu/~dbamman/fil... Some things to call out: 1/7

🚨 [Token][ization] Paper Alert 🚨 Tokenization is a wildly understudied area of language modeling despite it having effects across all of NLP. Over the past ~8 months, 32 (!) tokenizer researchers put together the most comprehensive survey of the field. Check it out!

We can put a 10ft human skeleton in the front yard of anyone who wants one. Why don’t we have the supply chain to put a T. rex or triceratops skeleton on every corner?

If you're at #COLM2026, come check out our poster for "Spoiler Alert" (arxiv.org/abs/2604.09854). We ask why LLM fiction is so bad at holding narrative tension, and create a metric to measure tension in short stories. What improves LLM fiction on this metric, it turns out, is narrative planning.

Delivering dinosaurs for exhibit at the Boston Museum of Science. Arthur Pollock, 1984

Cool post-doc opp for recent graduates in computer science & AI from Schmidt Sciences; broad range of AI topics that intersect with comp social sci. & humanities, esp. problems with alignment, interpretability, and knowledge production. Great foundation to work with: jobs.lever.co/schmidt-enti...

I'm recruiting 1-2 PhD students to join our lab in Fall 2027, through either Computer Science or Information Science at the University of Colorado Boulder! Looking for people with interests in NLP plus [healthcare | literary studies | narratives | social media | etc.]. Join us! 🏔️☀️

[New preprint] How do images in VLMs align with words? In this work, we found a set of attention heads responsible for OCR. But to our surprise, these heads were actually able to verbalize much more than just text. So, we used them to create a simple logit lens for image tokens!

🎙 In our new episode, we spoke with @veredshwartz.bsky.social (Assistant Professor of Computer Science at The University of British Columbia) and highlighted her book Lost in Automatic Translation: Navigating Life in English in the Age of Language Technologies. (1/7🧵)

Great thread, great blog and report. And since we're plugging past work that is newly relevant, my heart has always been with this 2023 one: muse.jhu.edu/article/898331

Paper: arxiv.org/abs/2609.23178 Code: github.com/Historical-A... Team: Ted Underwood, Ziliang Qiu, Laura K. Nelson @lauraknelson.bsky.social, Sarah Griebel @sgriebel.bsky.social, Edwin Roland @teddyroland.bsky.social, Wenyi Shang @wenyishang.bsky.social , Matthew Wilkens @mattwilkens.bsky.social.

We are hiring! ✨ Come work with me at @uvadatascience.bsky.social! We are recruiting an Open Rank Tenured/Tenure Track Prof of Data Science with a background in social science, legal, policy, STEM or humanities disciplines to join our Data and Society area. Apply here: apply.interfolio.com/191214

arxiv.org/abs/2510.08831 updated version of our article from last year @wouterhaverals.bsky.social and I updated the title & experiments. Check out this strengthened version, esp. if you are thinking abt cultural evaluation & benchmarks. @arnicas.bsky.social @dmimno.bsky.social @mariaa.bsky.social

UW iSchool is also hiring for a tenure-track Assistant Prof. in AI, broadly defined, and for two positions in Information Systems and Management. TT AI: apply.interfolio.com/192128 TT Info Systems & Info Management: apply.interfolio.com/192089 Teaching-track IS & IM: apply.interfolio.com/192126

We are hiring for several positions in the UW Information School, including a tenure-track position in Library and Information Science. Please share with anyone interested in these areas! Deadline October 30. Job: apply.interfolio.com/192127

while we continue to do battle with bot swarms (and hope to agreed on a more advanced strategy with Tufts), this paper appeared. Useful as the Perseus DL can be as a website, our bigger contribution is probably providing data for many other projects. This only begins the story.

and here's the link to the NLP class David @dasmiq.bsky.social and I co-taught last fall. hopefully it will be helpful to people as well! siwu.io/nlp-class/

2nd Exchange Meeting of SCOOP: International Network for Automated Text Recognition of Historical Sources starts on Monday. Keynotes by Thibault Clérice, Anna Dolganov and David Smith will be streamed via Zoom. Check the website for the links.

What does the history of copying machines teach us? That we should leave copying to machines and focus on interpretation, intent, and meaning - our human powers.

Truth-making institutions producing official interpretations of specific situations through eg divination or ordeals are very similar across cultures. @hugoreasoning.bsky.social & Boyer’s fascinating evolutionary account of their emergence and success: buff.ly/XKxt8wl

Paper: arxiv.org/abs/2608.29530 We consider a variety of neural networks that perform seemingly-symbolic tasks, from small-scale models trained on simple list-manipulation tasks to LLMs performing tasks in math, logic, coding, and language. 2/n