Boxes in an archive

Project: Artificial Intelligence as a risk and opportunity for the authenticity of archives

The purpose of this project is to conceptualise and guide the design of AI based methods which embody and address archival professional imperatives by using open and available datasets from the Swedish National Archives (Riksarkivet) and the Swedish National Heritage Board (RiksantikvarieƤmbetet).

Image at the top: Anton Grabolle / Better Images of AI / Classification Cupboard / CC-BY 4.0

Project information

Principal investigators
Koraljka Golub, Linnaeus University
Anna Foka, Uppsala University

Co-investigators
Johannes Widegren, Linnaeus University
Ahmad Kamal, Linnaeus University
Marcelo Milrad, Linnaeus University
Saara Taalas, Linnaeus University
Isto Huvila, Uppsala University
Larissa von Bychelberg, Uppsala University

Participating organizations
Linnaeus University
Uppsala University

Financier
Wallenberg Foundations’ research program WASP-HS (Wallenberg AI, Autonomous Systems and Software Program – Humanity and Society)

Timetable
15 August 2023 to 14 August 2028 (extended to June 2029 due to parental leave)

Website at Uppsala University
https://www.abm.uu.se/forskning/pagaende-forskningsprojekt/ai-authenticity/

More about the project

The purpose of this project is to conceptualise and guide the design of AI based methods which embody and address archival professional imperatives by using open and available datasets from the Swedish National Archives (Riksarkivet, thereon RA) and the Swedish National Heritage Board (Riksantikvarieämbetet, thereon RAÄ).

The project also aims to:

  1. articulate Swedish archival imperatives (e.g., archival diplomatics, automatic classification and community assessment)
  2. conceptualise and guide the design of AI systems to support imperatives
  3. provide an evaluative framework to assess AI implementation and performance including societal impact, providing means to maintain and assess the authenticity of human memory.

This project will fulfill a series of objectives which include matching archival imperatives with AI possibilities by surveying AI implementations and technologies, and archival principles and problematics, including diplomatics and service objectives by creating an archival AI Conceptual Framework (CF).

Moreover, other objectives include formulating an evaluative methodology to assess the performance of the CF; exploring how archival AI implementations would address wider societal issues, such as: aligning archives to heritage futures (service objectives) and addressing issues of fake news and propagandistic/inauthentic cultural appropriations (diplomatics).

Research questions
In order to achieve the ROs above the project addresses two sets of research questions:

RQ1 – What would AI-based approaches on archival principles look like? What possibilities do they suggest about humane AI development? Do they counter the “black box” character of AI?

RQ2 – What benefits would an archival AI offer an institution and society? Could it help future-proof cultural heritage institutions, i.e., foster heritage futures? Can an archival AI, outside the archival setting, serve public interest (e.g., combating fake news and propaganda)?

Expected outcomes

  • A major outcome with international and societal impact comprises a tested framework for maintaining and assessing the authenticity of historical evidence to implement in archival and related institutions that would allow a safe and effective implementation of selected AI techniques.

  • Through collaboration with RA, related WASP-HS projects, VR’s HUMINFRA infrastructure, DARIAH-EU and InterPARES Trust AI partnerships, the project will further devise measures to implement the resulting framework and guidelines at national and international levels to maximise impact and adoption of safe and effective AI in archives and related institutions.

  • The creation of a national and international network of actors working in this field in collaboration with DARIAH-EU (www.dariah.eu) that both Linnaeus and Uppsala are members of.

Concrete outputs:

  • Two PhD theses
  • At least eight peer-reviewed academic publications
  • At least four conference presentations

The project is part of the research in:

Digital Humanities
Digital Transformation Knowledge Environment 
IInstitute

Advisory Board