David Bamman

Timeline

1. 1998–2006

   Classics and applied linguistics

   He earns a BA in classics at the University of Wisconsin–Madison in 1998 and an MA in applied linguistics at Boston University in 2006.

2. By 2015

   Research at the Perseus Project

   By 2015, he works as a senior researcher at the Perseus Project, Tufts University’s digital library of Greek and Roman history, literature and culture. His work there includes natural language processing for Latin and Greek, treebank construction and word sense disambiguation.

3. 2015

   A doctorate and Berkeley appointment

   He completes a PhD in computer science at Carnegie Mellon University’s School of Computer Science, Language Technologies Institute, in 2015. In 2015, he joins UC Berkeley’s School of Information as an assistant professor.

4. 2019

   A dataset of literary entities

   In 2019, he coauthors “An annotated dataset of literary entities” with Sejal Popat and Sheng Shen. The dataset contains 210,532 tokens from 100 English-language literary texts, labeled for entities such as people, places and organizations.

5. 2020

   An NSF CAREER award

   He receives an NSF CAREER award in 2020 for research on computational methods that improve natural language processing for fiction. The project examines how common-sense knowledge in literature can improve information systems and supports teaching these methods to students from the humanities and social sciences.

6. September 2020

   Latin BERT

   In September 2020, he and Patrick J. Burns present Latin BERT, a language model trained on 642.7 million Latin words. Their case studies cover part-of-speech tagging, predicting missing text, word sense disambiguation and semantic search.

7. 2021

   Gender bias in GPT-3 stories

   With Li Lucy, he coauthors “Gender and Representation Bias in GPT-3 Generated Stories” in 2021. Using topic models and word similarity measures, they find that generated stories reproduce gender stereotypes, including associations between feminine characters, family and appearance.

8. 2023

   Books memorized by language models

   In 2023, he coauthors “Speak, Memory: An Archaeology of Books Known to ChatGPT/GPT-4.” The study finds that these models memorize copyrighted books and perform better on memorized books in subsequent tasks, which affects the validity of their use in cultural analysis.

9. September 2026

   Camera distance across film history

   In September 2026, he coauthors “Measuring the evolution of camera distance across a century of film.” The study uses computer vision and AI to analyze camera distance in 5,205 films shown in theaters between 1922 and 2025, examining changes associated with sound, television and gender representation.

10. As of October 2026

    Associate professor at Berkeley

    As of October 2026, he is an associate professor in UC Berkeley’s School of Information, with an affiliated associate professor appointment in Electrical Engineering and Computer Sciences and affiliated appointments in Linguistics and Computational Precision Health. He is a faculty member of the Berkeley AI Research Lab, a senior fellow at the Berkeley Institute for Data Science and a faculty leader of its Cultural Analytics Group. His research applies natural language processing and AI to questions in the humanities and social sciences.