Lecture
Toward an AI Attuned to Dissent and Consensus in Historical Events: Evidence from Wikipedia
- Date
- Friday 12 December 2025
- Time
- Address
-
Johan Huizinga
Doelensteeg 16
2311 VL Leiden - Room
- Digital Lab, Huizinga 0.09 (in person, registration not necessary)
Toward an AI Attuned to Dissent and Consensus in Historical Events: Evidence from Wikipedia
This talk presents the “Dissentometer,” a project that applies heatmaps and clustering to the stories about the past recorded in Wikipedia’s 365 languages. By mapping the interplay of consensus and dissent between languages, the method exposes intercultural disagreement and agreement. The Dissentometer makes these dynamics visible at a global scale, yielding an account of worldwide history-telling at resolutions previously unattainable.
At the same time, these insights point to a role for the humanities and social sciences in the future of AI. As machines increasingly summarize, classify, and interact with human knowledge, designing machines that can distinguish informed dissent from pseudohistory becomes critical for designing accountable systems.
Biography: Jo Guldi, historian and data scientist, is professor of Data and Decision Sciences at Emory University, where she founded and co-directs the Center for the Future of Trust, an interdisciplinary lab whose projects sit at the intersection of history, artificial intelligence, and public reasoning. Guldi’s projects include the Dissentometer, a project for providing accurate AI benchmarks that measures patterns of dissent and consensus across historical and digital corpora; Democracy Viewer, a no-code platform for analyzing change in political speech; and Text Mining the Documentation of Climate Change, which applies natural language processing to trace evolving narratives about environment, policy, and responsibility.
A former junior fellow at Harvard and member of the Brown History faculty, and author of award-winning books like The Long Land War and The History Manifesto, Guldi’s current research embodies not merely interdisciplinary but also hybrid scholarship, where computer science and historical research advance simultaneously. Her research has been covered by The Atlantic Monthly, The Wall Street Journal, The Papers of the National Academy of Science, The BBC, CBC, The Dig Podcast, NovaraFM Media, and many other social media and news outlets.