From Paper to FAIR

Using AI to Digitise and Curate Sensitive Research Data

Abstract

This paper examines a case study on the digitization and "FAIRification" of Erling Birkedal’s longitudinal surveys on religiosity (1994–2017). It details a technical workflow involving AI-assisted transcription of handwritten data, and subsequent curation using AI and OpenRefine. The study evaluates the legal framework for this process, specifically the principle of "effective anonymization", while addressing the environmental footprint of AI infrastructure. Findings indicate that AI chatbots reduce the labor required to transform analog legacy data into structured formats, provided that a "human-in-the-loop" verification process is maintained to mitigate hallucinations. Furthermore, the study suggests that AI empowers librarians to perform complex data engineering tasks without deep programming knowledge, effectively lowering the entry barrier for advanced data stewardship. The article concludes that AI can be a useful tool in research data management of sensitive data, but its responsible implementation requires expert oversight to ensure both data accuracy and ethical compliance.