As artificial intelligence (AI) transforms how we work and share information, archivists may find themselves engaging with AI in unexpected—and perhaps unsettling—ways. This presentation details a project launched in 2024 at Rensselaer Polytechnic Institute (RPI) in Troy, New York. The RPI president created the Project Bridge Generative AI team and appointed the institute’s archivist to lead research for drafting a commencement speech for a woman who died in 1903. The project was a major success and drew attention from thousands, but it also left archivists reflecting on ethical and environmental concerns.
RPI archivists delved into the world of prompt engineering, large language models, and primary source research to train GPT-4 to generate text in the style and voice of Emily Warren Roebling (d. 1903), wife of the chief engineer of the Brooklyn Bridge. The project was initiated to award her a posthumous honorary degree during RPI’s bicentennial ceremony. While honored by the president’s trust and excited by the archive’s visibility and collaboration with AI experts, concerns eventually emerged about AI’s environmental impact and the new ethical imperatives it raises for archivists.
As stewards of the historical record, archivists often hold the power to appraise and decide who is remembered and why. However, the use of AI raises new dilemmas and urgent questions:
How do we ensure we apply the same caution when using machine learning to interpret human experience?
What right do we have to consume finite natural resources in the process of using AI?
Though we are constantly encouraged to adopt new technologies, professionals must recognize that generative AI is resource-intensive: it requires raw material extraction, consumes large amounts of electricity, generates toxic waste (such as mercury and lead), and uses water to cool electronic components. Archivists must openly debate this new paradigm and become advocates for the sustainable use of AI.
This presentation will explore the many achievements and limitations of using AI in archival projects in the short term and invite the audience to reflect on the need for greater ethical and environmental awareness in the use of these technologies.