More broadly, and in conversation with Laura Nelson’s recent work, the goal isn’t “unbiased AI.” It’s developing better ways of reasoning under conditions where evidence is mediated, constructed, and otherwise severed from the world.
Breen is right that "we truly have no idea."
Also, this sprawling conversation is not limited to the humanities + CS. Sociologists have been very active. See Kozlowski et al. "In Silico Sociology" — and many critiques of "silicon sampling" — and this interesting theoretical piece by Skarpelis:
Also, Anna Skarpelis has been writing about this question: bsky.app/profile/akms...
Anna Skarpelis
Ted Underwood
Ted Underwood
📖 New publication out 📖
This piece is a first attempt at articulating why computational social science needs qualitative and historical approaches—especially as we begin to rely more heavily on simulation.
sociologica.unibo.it/article/view...
📖 New publication out 📖
This piece is a first attempt at articulating why computational social science needs qualitative and historical approaches—especially as we begin to rely more heavily on simulation.
sociologica.unibo.it/article/view...
Interesting take on "historical LLMs" as potential research tool for the humanities. I have been working on a paper for SSHA in November 2026 that will address some of these issues and am curious to follow what people publish in the interim
It’s part of a symposium engaging the one and only Carlo Ginzburg (read The Cheese and the Worms if you haven’t yet 🧀🪱).The core argument: Qualitative and historical work doesn’t just “contextualize” AI,it gives us a toolkit for thinking critically about what counts as knowledge in the first place.
Many thanks to Lorenzo Sabetta, @giozzam.bsky.social and @filippobarbera.bsky.social for the invitation. And more to come—a follow-up piece will be presented at SSHA this fall in a panel organized by @profyangzhang.bsky.social and Simeon Newman
🔗 Read the article (open access):
sociologica.unibo.it/article/view...
🔗 Explore the full issue, with contributions from Peter Bearman, Alix Rule, Lorenzo Sabetta, Mario Cardano, Maurizio Catino, Fabrizio Martire and Vincenzo Mele
sociologica.unibo.it/issue/view/1...
📖 New publication out 📖
This piece is a first attempt at articulating why computational social science needs qualitative and historical approaches—especially as we begin to rely more heavily on simulation.
sociologica.unibo.it/article/view...
