Author

Do large language models possess any kind of “identity”? On the one hand, such models ostensibly stand apart from the identity formations of human life; on the other, they are trained on texts and images derived from a world that understands identity – both claimed and attributed – as an inherent element in navigating social space. This article addresses how far this tension may undermine some of AI’s grander claims, but nonetheless offers distinct ways of thinking about our position in the world, and our relationship to history.

This article starts from a recent paper on the application of large language models to mathematical problems, to ask how far “evaluating” and “solving” are conceptually useful in understanding the past. It explores whether the idea of knowledge as “discovery” can be applied to history, and, if so, what that would mean for the parameters invoked by, and applied to, machine learning.

How do maps created fixed points, and what do those points reveal about history? This article explores how the information provided by maps can be rendered flexible and responsive throughAI-generated models, and so open up new avenues of historical research. 

The intriguing confluence of machine learning and history, where trust plays a central role. With its historical significance in both medieval and modern contexts influences our understanding of the past. The article discovers the challenges and far-reaching implications of trusting machine learning as effective arbiters of history.