ACA 2023 In Person Pre-Conference Workshop: “Artificial Intelligence and Archives: the I Trust AI Research Project.

ACA 2023 - Belonging - Considering archival bonds and disconnects

Pre-Conference Workshop

“Artificial Intelligence and Archives: the I Trust AI Research Project.”

When: June 28, 2023, Delta PEI Charlottetown

Time: 1:00 PM to 4:00 PM AT

Capacity: 50 participants

The ongoing availability, accessibility, and trustworthiness of public records is under threat, due to sociotechnical changes that have left our public institutions staggering under an overwhelming abundance, complexity, and lack of control of records. Without automated means of processing, we will increasingly lose control of public records, and ultimately, the public will lose trust in archives. However, implementing automated solutions - especially Artificial Intelligence solutions - raises a whole new set of costs and risks, from the expenses involved in implementing systems, to the risks of privacy breach, bias and discrimination, and loss of discretion.

This workshop will explain the fundamental ideas on which AI is based, discuss types of AI tools available, and identify both the advantages and issues presented by their use on records and archives. It will then outline the way the new InterPARES project, I Trust AI, aims to determine how the records and archives community can develop its own AI tools based on archival principles, and leverage them to support society's demand for trustworthy public records. Finally, it will present some of the studies that are being carried out by the I Trust AI researchers, and their outcomes to date, and will involve the workshop participants in the analysis and discussion of such outcomes.

• Muhammad Abdul Mageed and Peter Sullivan will explain types of AI systems and tools.
• Luciana Duranti will present what kind of tools are needed by the profession vs those that have been used to date, and the reasons why the 5th phase of the InterPARES project will be necessary.
• Corinne Rogers will discuss the structure of the I Trust AI project and, with Peter Sullivan, provide an overview of the studies in progress
• Tracey Lauriault will discuss two I Trust AI case studies about the preservation of digital twins and smart grids
• The participants will be divided in 4 groups, each chaired by one of the speakers, to discuss the studies and propose additional studies
• The participants will rejoin to discuss the new proposals

Learning Outcomes

The participants will learn to:

• recognize different types of AI tools
• identify the type of tool needed for different archival functions and activities
• assess the conditions required for the use of AI tools (e.g. amount of terabytes of digitized material)
• understand what is feasible and what is not
• learn what current research is being done on the use of AI in archives

Muhammad Abdul-Mageed is Canada Research Chair in Natural Language Processing and Machine Learning. His research focuses on deep representation learning and natural language socio-pragmatics, with two main goals: (1) development of `social’ machines for improved human health, safer social networking, and reduced information overload; and (2) use of machine learning as a vehicle for making discoveries with and about human language. He is currently leading several funded projects, including InterPARES Trust AI, and has been funded by SSHRC, NSERC, CFI, Google, Amazon, AMD, among others.

Luciana Duranti is a Professor at the UBC School of Information, where she teaches archival diplomatics, appraisal, preservation of digital records, and history of recordkeeping. Her research aims to find solutions to digital record issues that can be universally applied. Since 1998, she has been the Director of InterPARES (International Research on Permanent Authentic Records in Electronic Systems), a multi-national and multi-disciplinary research project studying the long-term preservation of authentic electronic records, which is now in its 5th phase. She has been the lead investigator also for other SSHRC-funded projects, such as Records in the Clouds and Digital Records Forensics.

Corinne Rogers is the Project Coordinator for InterPARES Trust AI (UBC, 2021-2026), a multidisciplinary, international partnership researching the uses and applicability of Artificial Intelligence in archival workflows to ensure trust and trustworthiness of records and data. She is an adjunct professor in the UBC School of Information, where she teaches diplomatics and digital preservation.  She was most recently a Systems Archivist at Artefactual Systems, lead developers and organizational home to open source projects for digital preservation, AccessToMemory (AtoM) and Archivematica.

Peter Sullivan is a Graduate Academic Assistant working under Luciana and Muhammad. His research focuses on Speech Processing, particularly Automatic Speech Recognition and classification (State Detection, Language and Dialect ID). Prior to returning to academia, he was a high-school Computer Science teacher, and brings with him his passion for demystifying complex technologies.

Tracey P. Lauriault is Associate Professor, Critical Media and Big Data, School of Journalism and Communications, Cross Appointed to Digital Humanities, and board member of the Institute for Data Science at Carleton University in Canada. She a founder of the field critical data studies, open data and Open Smart Cities. As a data and technological citizen, she examines large and small data and technology systems to make them more just, inclusive, equitable and environmentally sustainable. https://orcid.org/0000-0003-1847-2738