Led by Oxford scholar Pieter François, the World History Lab translates core historical questions—using AI and related tools—into a set of explanatory models that can be tested empirically. Its aim is to identify patterns in world history through quantitative methods over long timescales. The research team encodes evidence scattered across archaeological materials, chronicles, institutional histories, and records of religious rituals into traceable, auditable variable systems (which can reach thousands of dimensions). They then use statistical inference and computational modeling to evaluate the explanatory power of hypotheses about social complexity, institutional formation, war mobilization, and religious–ritual structures. Crucially, this approach does not reduce history to clean data; instead, it treats uncertainty, contestation, and gaps in the sources as part of the modeling process itself. Through explicit variable definitions, evidence-chain annotation, and reproducible coding protocols, it enables scholars to debate and verify findings within a shared analytical framework.
As an exemplary attempt for integrating AI into historical research, this case both sketches a pathway for AI-assisted cross-cultural comparative work and advances a new theoretical account of how historical data can be used to produce and test explanations.