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 digitization and born-digital records grow, manual archival processing becomes less feasible. This paper surveys research on AI and archives through the Records Continuum, finding 53 references for a representative overview rather than a systematic review. It identifies AI use across archival workflows and points to further work integrating AI into archival information systems and scholarly access.
Problem
Expanding digitized and born-digital collections make manual archival processing less feasible, while records still require quality and trust assessment.
Source: Introduction · pp. 1–2Method
The authors survey recent research on AI and archival processes, organizing the literature through the Records Continuum model.
Source: Introduction · pp. 2, 4, 5Results
The survey analyzes 53 references as a representative overview, not a systematic review, and identifies AI use across recordkeeping processes and four thematic areas.
Source: Introduction · p. 5, A Survey of Archives and AI · p. 14Takeaways & Limitations
Further work could better integrate AI into archival information systems and develop AI-powered ways for scholars to use archives.
Source: A Survey of Archives and AI · p. 13The survey uses AI primarily as a proxy for machine learning, while also including professional, cultural, and societal consequences of automated systems.
Source: Introduction · p. 4
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
As digitization expands archival collections, manual processes become less feasible, prompting greater use of AI across recordkeeping. The survey maps this work through the Records Continuum and presents a representative, not systematic, overview of 53 references.
Introduction
- As digitization and born-digital records expand, manual appraisal, selection, and review become less feasible, while archivists still need to assess records’ quality and trustworthiness.Source: Introduction · pp. 1–2
- The survey identifies AI use throughout archival processes, beyond earlier collection-specific experiments, and organizes its review around the Records Continuum.It also notes that no prior survey of AI and archives was known to the authors.Source: Introduction · p. 2
- The article clarifies its scope and methods, reviews literature in four thematic categories, and discusses trends and future opportunities.Source: Introduction · p. 2
- The survey takes an inclusive recordkeeping scope, covering records’ creation, management, and use, and uses the Records Continuum to structure its review.The model covers creation, capture, organising, and pluralising, with dimensions that can influence one another.Source: Introduction · pp. 2–4, Figure 1 · p. 3
- For this survey, AI primarily serves as a proxy for machine learning, while also encompassing automated systems’ professional, cultural, and societal consequences for recordkeeping.Source: Introduction · p. 4
- The authors searched selected archival and digital-humanities venues for 2015–2020 work, screened for explicit archival connections, then used citation lookup and expert-selected references to expand the set.They excluded purely technical studies without archival reflection and archival studies without explicit AI references.Source: Introduction · pp. 4–5
- The analysis covers 53 references as a representative overview, not a systematic review, and acknowledges the influence of the authors’ perspectives.Source: Introduction · p. 5
A Survey of Archives and AI
The survey uses the Records Continuum to map AI across archival theory and professional practice, automated recordkeeping, organising and accessing archives, and novel digital archives. It finds growing activity alongside unresolved questions about ethical scrutiny, lasting integration into archival systems, and the role of recordkeeping principles in AI.
Theoretical and professional considerations
- Digital transformation pressures archival concepts and professional roles, prompting calls to rethink appraisal, recordkeeping theory, and archivists’ skills.Debates include applying recordkeeping principles to algorithmic data and documenting explanations for algorithmic decisions.Source: A Survey of Archives and AI · pp. 5–6
Automating recordkeeping processes
- AI is being explored for appraisal, sensitivity review, and metadata creation, but successful use may require substantial preparation and workflow development.Reviewed approaches include classifiers trained to replicate expert email appraisal decisions and a tool combining archival frameworks with data-mining methods.Source: A Survey of Archives and AI · pp. 6–8
Organising and accessing archives
- AI-supported content extraction and indexing complement provenance and original order with searches based on record contents and entities.Examples include handwriting and optical character recognition, historical indexes, and entity-based indexation across archival fonds.Source: A Survey of Archives and AI · pp. 8–9
- Distant reading applies methods such as topic modelling and language models to identify topics, periods, and keywords in large collections.The surveyed work also considers indexing visual material and distant viewing.Source: A Survey of Archives and AI · p. 9
- Search and retrieval projects use classification, aggregated descriptions, metadata, and visualisation, while researchers call for more context-sensitive and critical search.Critical search includes alternating algorithm use with reading and reflection.Source: A Survey of Archives and AI · pp. 9–10
Novel forms of digital archives
- AI helps create and organise digital archives beyond established institutional collections, including sensor records, protest events, and web archives.The survey also discusses AI-supported efforts to surface archives maintained by marginalised communities and to use archival principles in curating training data.Source: A Survey of Archives and AI · pp. 10–12
Discussion and future directions
- The survey finds that most discussion of AI’s technical, ethical, and societal implications in archival processes remains to be done, while experiments still need to become lasting practice.The authors also identify gaps in interdisciplinary work, archivist training, and approaches for building trust in AI techniques.Source: A Survey of Archives and AI · pp. 12–14
Discussion
- Surveyed work concentrates on organising and pluralising records, with less on capture and minimal attention to creation; integrating this continuum into archival systems remains largely ahead.AI-based alternative orderings can complement traditional provenance and original order.Source: A Survey of Archives and AI · pp. 12–13
A APPENDIX
The appendix lists publication venues considered for the survey and identifies the latest issue or edition included for each.
A APPENDIX
- Table 1 lists the venues and the latest journal issue or conference/workshop edition considered for each.Source: Table 1 · p. 21
- Listed journal venues include Information Discovery and Delivery, the Journal of Contemporary Archival Studies, and the International Journal of Digital Curation and International Journal of Digital Humanities.Source: A APPENDIX · p. 22
- Other listed journals include the International Journal of Document Analysis and Recognition, International Journal of Humanities and Arts Computing, and International Journal of Information Management.Source: A APPENDIX · p. 22
- The list also includes the Journal of Archival Organization, Journal of Cultural Analytics, Journal of Data Mining and Digital Humanities, Journal of Information Science, and Journal of Open Humanities Data.Source: A APPENDIX · p. 22
- Further listed venues are JASIST, the Journal on Computing and Cultural Heritage, and conference or workshop entries including ICFHR and VIPERC.Source: A APPENDIX · p. 22
- The venue list also includes a 2020 conference/workshop entry.Source: A APPENDIX · p. 22