2026/05/12 by Heinrich Ammerer · 1 voice
Social Sciences · #Educator Training and Historical Pedagogy #Innovative Teaching Methodologies in Social Sciences #Computational and Text Analysis Methods
paper · doi:10.14324/herj.23.1.11
This exploratory study investigates the capacity of AI image generators to produce historically coherent visualisations and evaluates their didactic potential. In 2024, 1,024 images generated from six leading models were analysed using a standardised visual content analysis tool across four dimensions: historical significance, epochal authenticity, comparative lifeworld depiction and archetypal representation. The findings indicate that, while the output adequately depicts epochal atmospheres, it lacks concrete historical referents and exhibits poor independent historical judgement. Archetypal hero motifs predominantly emerged in medieval and classical guises, thereby reinforcing Western iconography. Only one model produced largely recognisable events and comprehensible comparisons through automated prompt augmentation, despite still reproducing biased historical narratives and stereotypes. The results highlight the potential of AI image generators to confront students with various conceptual problems, biases and anachronisms that can be discussed in class to encourage historical thinking. The study calls for didactic frameworks that integrate AI-generated images into structured historical enquiry to promote source criticism and reflective prompt design.