BibTeX

@incollection{broden-etal-2026-retouching-362272,
	title        = {Retouching and Refiguring lIterary Criticism: Experiments with a Generative Model for Analyzing Book Reviews},
	abstract     = {In the chapter, Daniel Brodén, Lina Samuelsson, and David Alfter ­ examines the use of GPT-4o to analyze Swedish literary criticism from 1905–1906. Drawing on defamiliarization and the notion of distant technology, and framing AI as artificial communication rather than intelligence, the chapter undertakes two linked experiments to enrich analysis of familiar material based on a previous study by Samuelsson (2013). First, ­ Brodén, Samuelsson, and Alfter address the challenge posed by the poor quality of digitized newspaper texts in the National Library of Sweden’s (­ Kungliga biblioteket, hereafter KB) collection, showing that the model’s output aligns more closely with the originals than does the noisy OCR, while arguing that it is better understood as a probabilistic retouching rather than a reconstruction, thereby introducing epistemological ­ un­ certainty. Second, using zero-shot prompting, they ask the model to identify discursive patterns and evaluative criteria in the reviews. The results suggest that GPT-4o can provide analytically meaningful perspectives, but its opacity creates methodological distance that calls for analytical caution, further re-readings, and reflection.},
	booktitle    = {Flows & Frictions: Mixed Methods for AI-Driven Research on Historical Media},
	author       = {Brodén, Daniel and Samuelsson, Lina and Alfter, David},
	year         = {2026},
	publisher    = {LIR Skifter},
	address      = {Göteborg},
	ISBN         = {978-91-89284-18-0},
}