QUESTIONS IN HISTORIOGRAPHY FROM THE NINETEENTH CENTURY TO THE AGE OF GENERATIVE AI

Abstract

History theory does not have a mature theory of questions. This reflects both historical and philosophical assumptions. As Holly Case has argued in The Age of Questions (2018), the big questions of the nineteenth century and their proposed final solutions arguably primed the murderous logic of genocide in the first half of the twentieth century. On her account, questions have become tamed as technical tools in historical monographs and reviews like this one. This picture of the twentieth century, though, runs up against R. G. Collingwood's historiographical logic of questions and the rise of erotetic logics in computer science. Computational erotetic logics have shaped the creation of large language models such as the GPT series and focused our attention on expressivity, effectivity, and classification in the relation of questions and answers. Collingwood's logic is different, using the relation of questions to questions to point to presuppositions. This metaphysical view of erotetic logic is timely, for it reminds why it might be so hard for historians to cut through with true propositions in an age of AI. Collingwood reminds us that a focus on truth-evaluable answers to questions does not explain why those questions were asked in the first place. Chasing chains of questions back to presuppositions, Collingwood argues that tackling what is assumed and what is lived with can help historians to change an unthinking world. In our age, this includes the idea of a shift from historians being the users of large language models to historians being the designers of new forms of relationship between people and information.