Project Profile Algorithmic Futures – How Large Language Models Translate History into Future Scenarios

The project aims to empirically investigate how Large Language Models (LLMs) translate historical texts—specifically late antique sources about “others” or “foreigners”—into future scenarios. This process of algorithmic futurization makes implicit historical narratives, cultural assumptions, and the epistemic roles of the models particularly visible. The project therefore guides students toward a reflective use of LLMs by allowing them to engage directly with the epistemic, philosophy-of-history, and ethical implications of these systems.

What happens when AI generates future scenarios from historical texts? Students examine this question through iterative analysis based on controlled prompt variations—differing in connotation, with and without context, and in both German and English. Using a structured analytical framework, they systematically examine and compare historical narratives, epistemic positions, and other features in the model outputs.

The course concept can be directly transferred to other disciplines in the humanities. Its methodology—historical texts --> LLMs --> future scenarios --> critical analysis—can be applied across different periods, text genres, and research questions. The course also introduces an LLM “stations” concept and incorporates an ethics-oriented dilemma pedagogy. Both elements can be adapted across disciplines to strengthen students’ responsible use of AI.