This paper aims to investigate the transformative role of artificial intelligence (AI) in digital preservation in higher education libraries (HEL) with a focus on integrating advanced technologies to overcome traditional challenges in knowledge management. The study presents a systematic analysis that encompasses process automation, metadata creation and enrichment and the protection of scientific research data. Initially, it demonstrates how natural language processing (NLP)-based tools enhance the automatic extraction of essential information, optimizing operational efficiency and ensuring the integrity of repositories. Then, it discusses the importance of metadata standardization – through international standards such as Dublin Core and PREMIS – to ensure interoperability between systems and promote global collaboration. The paper also highlights the relevance of AI in mitigating risks, such as format obsolescence and data corruption and in democratizing access to information, expanding digital inclusion. Finally, future directions are discussed that emphasize the need for modular tools, collaborative governance and robust ethical policies to consolidate the sustainability and adaptability of digital preservation solutions.
A systematic literature review was conducted in Web of Science and Scopus (2020–2025), using four iterative search expressions and predefined inclusion/exclusion criteria (e.g. open access, “information science library science”). After screening, 48 articles were included for qualitative, interpretative synthesis under a constructivist paradigm. The review maps AI applications across preservation workflows and standards (e.g. Dublin Core, Preservation Metadata Maintenance Activity – PREMIS, Metadata Encoding and Transmission Standard – METS).
The review identifies five consistent themes: automation of curation tasks via natural language processing (NLP) for scalable, more consistent metadata; semantic enrichment to improve discovery and reuse; predictive analytics for format obsolescence and proactive migration; interoperability gains through standards-driven metadata; and governance needs around transparency, bias mitigation and capacity-building. Benefits co-exist with constraints related to infrastructure, skills and ethical safeguards.
Results reflect the chosen timeframe, databases and filters and are limited to published, open-access literature in Library Information Science venues. No primary testing of tools was performed. Future work should include comparative case studies, shared evaluation metrics and longitudinal assessments of AI-enabled preservation.
Academic libraries can prioritize AI policies and governance, invest in modular and standards-aligned tools, develop staff competencies (AI literacy) and embed ethical review (transparency, audits, bias mitigation) to scale preservation, improve interoperability and free staff time for higher-value tasks.
However, as pointed out by (Subaveerapandiyan and Ugwulebo, 2024), barriers related to technological infrastructure and financial support still limit the ability of many institutions, especially in less developed contexts, to integrate advanced AI solutions. These limitations highlight the importance of investments in professional training and in the creation of adaptive tools that can be adjusted to the needs and constraints of each context. Despite the advances achieved, the application of AI in academic repositories is not without challenges, with ethical issues being a central concern. Transparency in the algorithms used for data curation is essential to avoid bias and ensure that technologies are inclusive and equitable. Hansson and Dahlgren (2022) warn of the risks of exclusion of underrepresented groups and the need for ethical guidelines that ensure equity in access to and use of scientific information. Furthermore, Sikhakhane and Mthombeni (2024) explore the legal and psychological implications of the use of AI in academic libraries, suggesting that clear regulatory frameworks are essential to protect the interests of users and ensure the accountability of institutions that implement these technologies. Issues, such as the protection of personal data and information security, also emerge as critical challenges, requiring a multidisciplinary approach to address the complexities associated with AI governance. Sustainable governance is also a crucial aspect in integrating AI into academic repositories. (Strecker et al. 2023) argue that long-term data preservation requires collaborative governance models that prioritize infrastructure resilience and adaptability to technological change. This view is complemented by Azeroual (2021), who reinforces the importance of data quality as an essential pillar for the ongoing integrity of digital collections. In this context, Bossaller and Million (2023)highlight the challenges associated with data management and the need to integrate modern solutions without compromising the reliability of information already stored. The future of AI in academic repositories presents several promising directions for future research. One priority area is the development of modular and adaptable AI tools that can be customized according to institutional and regional contexts. Bamgbose et al. (2024) suggest that these solutions can democratize the use of AI, enabling institutions with limited resources to implement advanced digital preservation practices. This approach could be explored in future research that evaluates how these modular tools can be designed to maximize their impact in different contexts. Another promising direction is the construction of global networks of academic repositories, which facilitate interoperability and promote the sharing of good practices. Moyo and Bangani (2023) argue that international partnerships are essential to connect repositories in a global ecosystem, expanding the reach and relevance of collections. Furthermore, Frank and Wylie (2023) highlight the role of semantic enrichment as a facilitator for interdisciplinary connections, allowing data stored in academic repositories to be used more broadly and impactfully. Finally, ethics must remain at the center of discussions about AI in academic repositories. Shal et al. (2024) emphasize that organizational leaders play a crucial role in creating policies that promote transparency, inclusion and social responsibility. Melero et al. (2023) complement this perspective by pointing out that the dissemination of scientific knowledge must be accompanied by practices that ensure equity in access to and use of information. The ethical and sustainable integration of AI is essential to ensure that the benefits of these technologies are widely distributed and that their impact is positive for science and society as a whole. AI has already proven to be an indispensable tool for transforming academic repositories.
