Artificial intelligence is coming for the archives. Colleges and universities across the country are investing heavily in AI research across the humanities and social sciences. Secondary and higher education classrooms are frantically adapting, or tossing out altogether, traditional pedagogical models to accommodate shifting expectations. At a more fundamental level, the integration of generative AI into digital search, on the one hand, and writing practices on the other, is rapidly changing the way people find, process, and communicate information in their daily lives.

Given these rapid developments, how might artificial intelligence change records management, archival research, and the future of historical scholarship and education? A recent History Lab in AHR 128, no. 3, explored AI, prompting R. Darrell Meadows and Joshua Sternfeld to offer a challenge to professionals: “To ponder the effects of artificial intelligence on the field of history quite often requires interrogating fundamental concepts such as truth, evidence, and authenticity.” This panel seeks to contribute to the conversations started in the AHR’s “Artificial Intelligence and the Practice of History” and address new developments since its publication.

In these early days, historians, archivists, and teachers have yet to form a collective response to these new uses. This panel seeks to offer a forum for continuing the discussions started previously in the journal and to offer the discipline ideas for how to handle the new terrain that has and will continue to emerge. The algorithmic turn in the humanities offers exciting promises but comes bundled with worrying perils that require analysis and open communication.

This panel will take a roundtable format with brief remarks by each contributor, followed by an open discussion with the audience. Panelists will explore AI from a myriad of approaches. Mark Humphries will articulate how archives might best use the technology to speed up digitization and open access to more users. Chris Deutsch will raise concerns about the environmental impacts of the technology since the server banks have to go somewhere. Drawing upon precedents in labor history, Henry Snow will propose a call for collective action that counters the narrative of inevitable progress employed by the commercial sector. Josh Sternfeld will discuss the need for “history red-teaming,” a systemic approach to addressing the historiographic and evidentiary reliability of large language models (LLMs) such as ChatGPT.

The panelists’ remarks, each in their own way, will confront the underlying question that accompanies changes in the discipline raised by new AI tools and systems: Are the traditional methodologies and theories of the dual disciplines of history and archival science on the verge of radical transformation, or will certain principles prove more enduring than ever? In what ways can the fields proactively contribute to technological developments ensuring proper preservation of and access to the historical record and historiography grounded in rigorous epistemological standards? To what extent should historians and archivists understand the complex functions of AI systems?