Syed, Sara Asad
Jaillant, L., & Caputo, A. (2022). Unlocking digital archives: cross-disciplinary perspectives on AI and born-digital data. AI & society, 37(3), 823-835.
Summary: This article explores the barriers to accessing born-digital archives in cultural institutions like libraries, museums, and archives. It highlights how most born-digital materials such as emails, web archives, and digital records are "dark" (inaccessible) due to privacy concerns, copyright restrictions, commercial sensitivities, and technical challenges. Using examples like the British Library's Wendy Cope email archive and the National Library of Scotland's Data Foundry, the authors argue that archives overly prioritize risk aversion, contrasting this with tech giants like Google that exploit data maximization. They propose solutions like secure online access systems, consortia modeled on HathiTrust for born-digital content, and AI machine learning applications for sensitivity review. However, the paper also warns of AI pitfalls, including biases, errors, "black box" opacity, and ethical issues like fairness and transparency. The conclusion emphasizes cross-disciplinary collaboration to make archives more accessible while upholding ethical standards.
Evaluation: This article correlates well with collection management, especially around balancing access, preservation, and ethical stewardship in digital environments. In collection development, we're taught to prioritize user needs and inclusivity, but the authors show how privacy and copyright laws can stifle that. It is frustrating to think of valuable resources like personal emails or web snapshots sitting unused because institutions fear lawsuits or damage to their reputation. Practical suggestions like building secure online platforms or AI-assisted sensitivity reviews, could transform how we manage born-digital collections. For instance, integrating machine learning tools could help with weeding and appraisal processes, making large-scale collections more feasible without overwhelming staff resources. However there are ethical concerns. Biases in training data could perpetuate underrepresentation of marginalized voices. Overall, it reinforces the need for librarians to collaborate with tech experts and advocate for policy changes. I'm inspired to explore AI ethics more in my own research, perhaps focusing on how small libraries could adopt these tools affordably. This piece is a great reminder that collection management in the digital age is not just about acquiring items but ensuring that they are ethically and equitably usable.
No comments:
Post a Comment