Ghosts and the machine: testing the use of Artificial Intelligence to deliver historical life course biographies from big data

A study testing the use of Artificial Intelligence (AI) to generate historical life course biographies from big data reveals both promise and limitations. While AI, specifically Generative Large Language Models, effectively converts tabulated records into fluent and factually correct narrative text—such as those describing convicts and free individuals on Norfolk Island during Australia's colonial period (1788–1814)—it also exhibits inherent biases, uncritical perspectives, and a tendency to summarize data rather than delve deeply. The best outputs require human intervention to fully capture the nuanced aspects of history and heritage research. This experiment underscores AI's potential in historical narrative generation while highlighting the need for careful oversight to mitigate its limitations.

McLean, M. A., Roberts, D. A., & Gibbs, M. (2024). Ghosts and the machine: testing the use of Artificial Intelligence to deliver historical life course biographies from big data. Historical Methods: A Journal of Quantitative and Interdisciplinary History, 57(3), 146–162. https://doi.org/10.1080/01615440.2024.2398455