Source-linked AI summary
Archives and AI: An Overview of Current Debates and Future Perspectives
Giovanni Colavizza, Tobias Blanke, Charles Jeurgens, Julia Noordegraaf
TL;DR
As archives become increasingly data-driven, established approaches for addressing AI-related archival challenges remain unsettled. This survey frames recent work through the Records Continuum model and identifies four major themes, while highlighting gaps in translating experiments into lasting, ethically grounded archival practice.
Problem
Archives face new AI-related challenges as they become larger data sets, but established approaches for addressing them remain unsettled.
Method
The survey organises literature on archives and AI into four themes using the Records Continuum model as its framework.
Results
The literature spans theoretical and professional issues, automated recordkeeping, archival organisation and access, and novel digital archives, with emphasis on Organise and Pluralise dimensions.
Takeaways & Limitations
Future work should turn AI experiments into lasting archival infrastructure while strengthening ethical frameworks and integrating recordkeeping principles into AI.
Takeaways & Limitations
The survey leaves out related areas including AI law and AI ethics, which it identifies as topics for future work.
Abstract
from arXiv · showhide
The digital transformation is turning archives, both old and new, into data. As a consequence, automation in the form of artificial intelligence techniques is increasingly applied both to scale traditional recordkeeping activities, and to experiment with novel ways to capture, organise and access records. We survey recent developments at the intersection of Artificial Intelligence and archival thinking and practice. Our overview of this growing body of literature is organised through the lenses of the Records Continuum model. We find four broad themes in the literature on archives and artificial intelligence: theoretical and professional considerations, the automation of recordkeeping processes, organising and accessing archives, and novel forms of digital archives. We conclude by underlining emerging trends and directions for future work, which include the application of recordkeeping principles to the very data and processes which power modern artificial intelligence, and a more structural, yet critically-aware, integration of artificial intelligence into archival systems and practice.
Introduction
The section references the Records Continuum model.
- Introduction: The Records Continuum model is identified as Figure 1.The figure is attributed to Upward (2018).
A Survey of Archives and AI
The survey uses the Records Continuum model to organise research on AI and archives into four thematic areas: theoretical and professional considerations, automated recordkeeping, organising and accessing archives, and novel digital archives. It finds growing practical experimentation alongside critical concerns about archival principles, AI limitations, and future interdisciplinary integration.
- Survey structure: The literature is organised into four themes using the Records Continuum model: theory and professionalism, automated recordkeeping, archival organisation and access, and novel digital archives.The framework also accommodates discussion of emerging trends.
- Automating recordkeeping processes: AI is being explored to automate archival workflows, especially appraisal, metadata creation, and the management of expanding volumes of unstructured records.Automation discussions predominantly concern the Records Continuum’s ‘Capture’ and ‘Pluralise’ dimensions.
- Organising and accessing archives: AI-based extraction, recognition, indexing, metadata synchronisation, and virtual collections are enabling archives to be organised and accessed in richer ways.Examples include Transkribus and the EHRI project’s combination of archival and user-generated metadata.
- Organising and accessing archives: Historians caution that AI search and retrieval should respect archival hierarchy and context rather than reinforce confirmation bias or obscure gaps and absences.The survey also identifies the need to address sensitive information while enabling freedom-of-information access.
- Novel forms of digital archives: AI can capture and organise records beyond institutional collection policies, including protest-related digital resources and born-digital archives.Automated crawling, classification, entity extraction, authentication, and unsupervised analysis can expose otherwise overlooked events, contributors, and patterns.
- Discussion and future directions: The survey finds activity concentrated on the Records Continuum’s ‘organise’ and ‘pluralise’ dimensions, while calling for more reflexive, critically aware, and interdisciplinary integration of AI and recordkeeping.It highlights limited reflection on AI’s limitations and unintended consequences, alongside the need for collaboration among archivists, humanists, and computer scientists.
A APPENDIX
The appendix lists the venues surveyed and records the latest volume, issue, year, or edition considered for each venue. It includes journals, conferences, and workshops spanning archival, information, digital humanities, and cultural heritage research.
- Venue coverage: Table 1 defines the venue list and records each venue’s latest issue or edition considered.For journals, this includes volume, issue, and year; for conferences and workshops, it records the latest edition.
- Journals: The listed journals include Information Discovery and Delivery, International Journal of Digital Curation, and International Journal of Digital Humanities.The listed latest issues are 48, 3, 2020; 15, 1, 2020; and 2, 2019, respectively.
- Journals: The appendix also covers document analysis, humanities computing, archival organization, information management, and cultural analytics journals.Examples include International Journal of Document Analysis and Recognition, International Journal of Humanities and Arts Computing, Journal of Archival Organization, International Journal of Information Management, and Journal of Cultural Analytics.
- Journals: Additional venues include journals focused on data mining, information science, open humanities data, information science and technology, computing and cultural heritage, and contemporary archival studies.The appendix records issues or editions from 2020 for these venues, except the Journal of Open Humanities Data entry, which lists volume 6, 2020.