Fragments, not prompts: five principles for writing history in the age of AI

Abstract

This article offers five principles for historical writing in the age of generative AI — principles shaped not by fear of machines but by a confrontation with their fluency. As tools like ChatGPT increasingly saturate education, research, and public discourse, historians must reckon with what these systems can and cannot do. They summarize but do not listen, reproduce but do not interpret, and excel at coherence but falter at contradiction. Through close engagement with five Holocaust testimonies recorded in 1995–from Poland to La Paz to Connecticut — this essay refuses narrative smoothness in favor of fracture, silence, and ethical entanglement. The stakes are not technical but epistemological: What becomes of history when it is made machine-readable? And what remains essential to the historian’s craft? Grounded in archival work with survivor testimony alongside AI outputs, I insist that historical writing is not an output but a risk — a form of presence, of dwelling, of making meaning where no prewritten frame will do. It is not about outperforming the machine but about sounding nothing like it.