New Ways of Understanding the Past Using Generative AI

From Hollywood to local classrooms, Generative AI seems to be everywhere these days and it’s always controversial. But can it help us unlock news ways of doing research, even in conventional disciplines like history? In this talk, historian Mark Humphries shares his research on fur trade families to show how artificial intelligence can be used to re-examine old historical problems in new ways. Going beyond programs like ChatGPT, Humphries explores (and demonstrates) how generative AI models from OpenAI, Anthropic, and Google actually work behind the scenes and reflects on their often overlooked strengths and sometimes overemphasized limitations. He argues that although much of the media discussion is focused on “chatbots” the most revolutionary aspects of the technology have little to do with these one-on-one, “conversational” interactions. Instead, by demonstrating how they can be deployed at scale to speed up the research process to examine thousands of handwritten documents in only a few minutes, Humphries invites us to consider how AI may change the face of history and the social sciences more broadly in the months and years to come.