Libraries, archives, and records institutions play a vital role in shaping society’s relationship with the past. As artificial intelligence (AI) technologies continue to evolve, they are transforming the way these institutions operate and impacting workflows, priorities, and professional practices. This article explores how AI is influencing the sector, highlighting both emerging opportunities and the challenges for the future of cultural heritage management.
The Bright Side
New technologies are increasingly integrated into core functions across libraries and archives, reshaping traditional workflows and enhancing service delivery. From digitisation to collection development, these tools are enabling institutions to work more efficiently and make their collections more accessible and discoverable. Let’s take a look at some of the ways this is happening.
AI-enhanced technologies are revolutionizing the process of archival digitisation by automating and improving tasks that were once time-consuming and labor-intensive. Optical Character Recognition (OCR), for example, uses machine learning algorithms to accurately convert scanned documents and images into searchable, editable text, greatly accelerating access to archival materials. Predictive preservation leverages AI to analyze environmental data and usage patterns, allowing institutions to anticipate and prevent deterioration before it occurs, thereby extending the lifespan of physical and digital collections. Meanwhile, digital restoration employs sophisticated AI techniques to repair and enhance damaged or degraded materials, restoring them to a state closer to their original condition without the need for extensive manual intervention. Together, these AI-driven tools streamline workflows, reduce the burden on staff, and enable archives to manage their collections more effectively and efficiently.
The project “Living with Machines“, developed in partnership between The Alan Turing Institute, the British Library, and the Universities of Cambridge, East Anglia, Exeter, and London (QMUL, King’s College), explores the impact of big data on historical knowledge by interrogating the archive with the aid of data science and machine learning. In connecting these domains, the project realises an unprecedented scale of historical analysis. Below are some of these elements.
- New software tools were developed to refine the digitisation and access infrastructure of the British Library’s large database of nineteenth century newspapers.
- New computational tools were developed to analyse digitised map collections and connect information from maps to place names in newspapers and census records.
- Language models were developed from crowdsourced and computational models to answer research questions.
- Algorithms were developed to extract census micro data that enabled researchers to follow individuals from one census to the next.
- Datasets were combined to highlight social change occurring at different locations over time.
The Dark Side
While AI integration has swelled the repository of digitised records and increased the scale of analysis, this has not been without challenges. The issues that have generated the most concern and debate relate to data and interpretation bias, oversimplification, misuse of sensitive information, and the opacity of the “black-box” model.
While these models are still developing, AI integration can be prone to bias in interpretation, metadata tagging, and training datasets. The interpretation bias occurs when AI enabled systems reinforce dominant narratives or contribute to the underrepresentation of marginalised groups. This can be a direct inference of limited and/or unrepresentative datasets used to train the AI. Bias can also lead to an oversimplification in metadata tagging categories, erasing cultural and social specificity.
AI systems are trained using machine learning models. The lack of transparency and understanding in how these models work and how decisions are made, sometimes referred to as “black-box” decision making, can erode trust in the accuracy and reliability of AI derived information. Underlying this, is also the real possibility for the misuse, misidentification and misinterpretation of sensitive information, some of which was never intended to be used for AI training or content generation.
Revisiting Ethics
As artificial intelligence continues to evolve, its influence on memory institutions is becoming increasingly visible—reshaping workflows, redefining professional roles, and challenging long-standing assumptions about how we manage, preserve, and provide access to information. While AI presents significant opportunities for innovation and efficiency, it also raises critical questions around ethics, equity, transparency, and the preservation of human judgment in cultural stewardship. For libraries, archives, and records professionals, staying informed and actively engaged in the development and implementation of these technologies is essential. By doing so, the sector can help ensure that AI is used not only to enhance operations, but also to uphold the core values of access, accountability, and public trust.
Suggestions for further reading
Colavizza, G., Blanke, T., Jeurgens, C., & Noordegraaf, J. (2021). Archives and AI: An overview of current debates and future perspectives. ACM Journal on Computing and Cultural Heritage (JOCCH), 15(1), 1-15.
Ferro, S., Pelillo, M., & Traviglia, A. (2023). AI-assisted digitalisation of historical documents. The international archives of the photogrammetry, remote sensing and spatial information sciences, 48, 557-562.
Jaillant, L., & Rees, A. (2023). Applying AI to digital archives: trust, collaboration and shared professional ethics. Digital Scholarship in the Humanities, 38(2), 571-585.
Jaillant, L., & Zhao, L. (2025). Introduction: When data turns into archives: making digital records more accessible with AI. AI & SOCIETY, 1-5.
