Digitization of Tomorrow's Archives : Navigating the Opportunities and Risks of AI-Driven workflows

Sammanfattning

This master’s thesis examines how Artificial Intelligence (AI) may support digital workflows in small- to medium-sized archives, particularly in relation to repetitive and time-consuming tasks while maintaining professional oversight and historical accountability. The study adopts an exploratory qualitative approach, combining an introductory research review, along with semi-structured interviews with archival professionals and observational studies of archival workflows. The findings revealed fragmented digitization processes, limited staffing and large volumes of non-digitized material, creating both opportunities and challenges for a possible AI adoption. The study identifies areas where AI tools could relive repetitive tasks such as metadata generation, classification and transcription. At the same time, archival professionals express concerns related to reliability, traceability, ethical responsibility and the preservation of human expertise. In response to this, the thesis develops a modular and value-driven framework intended to support gradual and responsible AI implementation in archival environments. The proposed framework emphasizes human oversight, traceability, confidence thresholds, feedback loops and clearly defined exit strategies. Grounded in the theories of Human-Centered AI (HCAI) and Value Sensitive Design (VSD), the framework positions AI as a supportive tool rather than as a fully autonomous system. A preliminary practitioner review suggests that the framework may have practical relevance within archival contexts, while also highlighting challenges related to institutional readiness, technical competence, long-term maintenance and resource allocation. The assessment further emphasizes the importance of organizational trust, policy development and continuous staff involvement during implementation processes. The proposed framework contributes a practical and context-sensitive foundation for discussions surrounding responsible AI adoption in the cultural heritage sector. By addressing both technological possibilities and organizational realities, the study provides a basis for future long-term implementation research.