EAGER: Investigating and Training Digital Archive Communities in the Uses of Generative AI and the National AI Research Resources

The people who document, archive, and tell the stories of current and past cultural practices play an important part in societies, including how people view the past and possible futures. The growing omnipresence of generative artificial intelligence (AI) systems in society holds important implications for archival practices going into the future, making it imperative for those who do this essential cultural and archival work to understand and adequately use those same AI systems. This work connects AI and computing scholars with scholars who work in the study of digital archives and culture for mutual training and co-creation of new AI-based digital systems. As a part of this work, scholars learn and use resources provided by the National Artificial Intelligence Research Resource (NAIRR) during workshops to re-think how AI systems can and should be created to interact with specific cultural and archival practices. This research involves the exploration of current dominant uses of AI (especially generative AI) systems for certain digital archives and the development of new designs and early prototypes of possible systems based on the needs of those working on those digital archives. The research is complemented with the important educational aspect of training experts in archives and cultural practices on generative AI systems and the use of generative AI systems with NAIRR resources. This project connects researchers from information science and computer science and engineering with a community of archival and cultural practice researchers. As a part of this project, researchers work together to create contextualized and relevant AI systems using resources provided by NAIRR, with a particular emphasis placed on using a new multi-level framework for ethical engineering of AI systems. This work involves understanding impacts of behavior across individual, organizational, and social levels as AI systems are engineered. The deeper understanding of AI systems allows more nuanced considerations of culturally contextualized digital archives so that current archives and ways of knowing amongst relevant communities can be extended and augmented, as well as to understand what the development and integration of generative AI systems mean for the future of those digital archives. This use of contextualized digital archives also gives a unique opportunity to probe existing social knowledge structures present within generative AI systems. This research integrates several disciplinary perspectives, especially those from AI, computational cognitive science, and archival studies to study potential interactions between generative AI systems and certain digital archives across levels of behavior and analysis. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.