Uses machine learning to study literary imagination, and vice-versa. Likely to share news about AI & computational social science / Sozialwissenschaft / ็คพไผ็งๅญฆ
Information Sciences and English, UIUC. Distant Horizons (Chicago, 2019). tedunderwood.com
Ted Underwood
Loading...
For fans of Talkie-1930 and all the methodological questions raised by historical models, here's a new entrant into the field, TypewriterLM, trained up to 1913. Corpus, instruction-tuning datasets, and event dataset are released. arxiv.org/abs/2606.02991
We introduce TypewriterLM, a 7.24B History language model (LM) trained exclusively on English text predating 1913. Developing History LMs requires addressing challenges in data quality and availabilit...
was reading this and enjoying it โ and then enjoyed it more when it turned out to be a paper by Sil Hamilton and @dmimno.bsky.social
Ted Underwood
AI generated stories, aka the Elias Thorne literary universe www.404media.co/elias-thorne...
LLMs including ChatGPT, Gemini and Claude are obsessed with telling stories about lighthouse keepers and clockmakers, and one character named 'Elias Thorne' has made his way from chatbots to Amazon bo...
Say what you will about Urbana-Champaign, but in no other town/city that I have lived in have I seen a person with a full bumper decal on their car reading "See You, Space Cowboy..."
Possibly the key to everything.
www.nytimes.com/2026/06/10/s...
New preprint!
We introduce a new benchmark, SciConBench, with 9.11k scientific questions derived from Cochrane Systematic Reviews.
We find evidence that frontier AI agents **cannot** synthesize scientific conclusions well.
A thread ๐งต
w/ @hayoungjung.bsky.social & others!
I wanted to see which locally-runnable models are most effective for cultural heritage languages, this page has comparisons of 7 languages: mimno.github.io/PhiloBench/ llama3.1 and mistral are both powerful and fast!