AI Agents for Historical Research: A New Skill Set for Historians?

A New Skill Set for Historians explores the potential for Large Language Model based intelligent agents to support historical research, and makes the case for introducing such agents formally into graduate programs.

The talk demonstrates a range of GPT-4o of intelligent agents designed by the speaker to support specific aspects of historical research, including their application to analyzing and charting CSV  files, performing named entity recogntion, and supporting and cleaning up HTR. The speaker goes under the hood of the models to show how they have been configured and equipped with internal knowledge to deliver specific functionality, and emphasis the innate linguistic and coding skills contained in the underlying LLM, whilst also recognising their limitations.

The talk looks at the potential for agents to transform the relationship of researchers in dealing with online archival metadata and other digital resources, and draws on hands-on experience working with senior historians and the National Archives to explore this.

Finally, the talk explores the potential for multi-agent/multi-player historical simuations to be integrated into graduate teaching and looks at the structure of a simulation of international investment in the seventeenth century.