EVENT DESCRIPTION

BIO:

William J. Turkel is Professor of History at the University of Western Ontario and internationally recognized for his innovative work in digital history. He uses machine learning, text mining, and computational techniques in his study of the histories of science, technology and environment, drawing on many decades of programming experience. He is the author of Spark from the Deep (Johns Hopkins, 2013), The Archive of Place (UBC, 2007) and the open access textbook Digital Research Methods with Mathematica (2nd ed 2019). His current research focuses on the use of generative AI in historiography, history and theory, and historical methods. Dr. Turkel is a member of the Rotman Institute of Philosophy and the College of New Scholars, Artists and Scientists of the Royal Society of Canada (2018-25).

ABSTRACT:

Collaboration with Generative Artificial Intelligence will rapidly reshape the nature of historical thinking in the 21st century. Because GenAI is capable of performing complex, nonroutine, and cognitive tasks through natural language prompting, it drastically reduces barriers for developing hybrid human-AI systems that can read and understand vast numbers of sources (albeit in a distinctly non-human way). At the same time, GenAI systems operate so that they always have some possibility of so-called hallucination. Hallucinations can be mitigated by the use of more structured techniques for representing, analyzing, and reasoning about sources. Here, I make three arguments. First, contemporary information environments, including our collective record of the past, require sophisticated computational systems to make sense of. Second, we will need to understand this as an adversarial process. Historians have long practiced source criticism to evaluate reliability and authenticity, and assessed the biases, gaps, and silences in their archives. But GenAI supports new forms of ambiguity, influence, denial, deception, and disinformation and all of these techniques have a historical dimension. Third, the traditional slow scholarship of individuals writing monographs at the time scale of decades has to be supplemented by larger teams of humans and AIs, collaborating to create an evidence-based usable past in real time.

Attendance will be free, but advance registration is requested. Please RSVP at historyrsvp@uwo.ca or scan the QR code on the poster.