AI as a Historical Lens: An Experiment in Periodization of Russia’s State Photography Archive with Neural Networks

Abstract Chronology is an important framing mechanism in history and changes significantly based on who defines historical eras. The area studies field has recently grappled with the need to decenter perspectives and reconsider the sources that scholars use. This article uses deep learning artificial intelligence methods to process 169,634 images from the Russian State Documentary Film and Photo Archive (RGAKFD), a major archive of photography in the region, as containing a statist chronological logic, one defined by political change in the center. By peering under the hood of the algorithm’s predictions, by thinking with the machine, it is possible to see patterns in the images that may not seem crucial to the human eye. Looking at RGAKFD as a potential source of data for AI raises parallels between algorithmic bias and the Moscow-centric bias of sources, while also providing opportunities to use such methods as a tool for exploratory research.