Strategic Priorities for AI Integration and Digital Transformation in Academic Libraries

Introduction

The integration of artificial intelligence into academic libraries has transformed their role as dynamic knowledge hubs in higher education, aligning with global trends in digital transformation. Technologies such as natural language processing, machine learning, automated chatbots, and big data analytics have revolutionized library operations, enhancing information retrieval, resource digitization, personalized user services, and more effective knowledge management (Adilovic, 2024; Gasparini, 2022; Van Thinh, 2026). These advancements streamline workflows, support interdisciplinary research, and position academic libraries as critical drivers of academic excellence and innovation.

As universities worldwide face unprecedented growth in scholarly resources, with over three million articles published annually in Scopus-indexed journals (Semantic Scholar, 2024), AI has become a strategic imperative for modernizing library systems and improving access to information. Leading institutions have pioneered AI adoption, setting important benchmarks for academic libraries worldwide (Padilla, 2019). For example, the University of Hong Kong has employed AI-driven analytics for collection management, the University of Toronto has leveraged machine learning to enhance search precision, and the Bodleian Libraries at Oxford University have prioritized digitization and automation in their Strategic Plan 2022–2027 (Bodleian Libraries, 2022).

Despite these promising developments, prior studies have largely focused on specific AI tools (e.g., chatbots or automation) or isolated aspects such as ethics and knowledge management, while comprehensive analyses of strategic priorities across institutions remain limited (Huang, 2023; Cox et al., 2023). Furthermore, there is a noticeable gap in the development of scalable and contextually adaptable frameworks, especially for resource-constrained academic libraries in developing countries.

Technical barriers significantly hinder widespread AI adoption. Legacy infrastructures incompatible with modern AI systems, the high computational requirements of large language models, and challenges related to interoperability with existing integrated library systems continue to present major obstacles. Ethical and human resource challenges further complicate the process. Issues such as algorithmic bias, user privacy, lack of transparency, and a shortage of AI-skilled professionals are frequently reported, with particular severity in developing countries like Iran (Oladokun et al., 2025; Cox et al., 2023; Rahmani et al., 2022).

A critical research gap persists: the lack of systematic, comprehensive frameworks for strategic AI integration in academic libraries that can bridge global best practices with diverse economic and cultural realities (Huang, 2023). While some studies have examined AI ethics (Cox et al., 2023) or knowledge management, few have analyzed the strategic plans of leading universities to identify prioritized components and derive adaptable implementation models. Additionally, the potential of advanced AI technologies such as large language models (LLMs) in library contexts remains underexplored.

This study addresses these gaps by conducting a mixed-methods analysis of publicly available strategic plans from the top 25 universities (Times Higher Education 2024 ranking). The selection of these institutions is based on their recognized leadership in research, innovation, and digital initiatives, providing valuable benchmarks for academic libraries worldwide. The study examines approximately 1,200 pages of strategic documents to map AI-related priorities, identify implementation barriers, and propose a scalable strategic framework with particular attention to resource-limited settings such as Iran.

The research is guided by three main objectives: (1) to identify and prioritize key AI and digital transformation components in leading university library strategies; (2) to assess technical, ethical, and resource-related barriers to AI adoption; and (3) to develop a scalable and adaptable framework for AI integration across diverse contexts. These objectives are pursued through the following research questions: (1) What AI-related components are prioritized in the strategic plans of top-ranked university libraries? (2) What are the main barriers hindering AI adoption in academic libraries? (3) How can global insights from leading institutions inform contextually appropriate AI strategies for libraries in developing countries?

By systematically mapping strategic priorities and proposing a practical framework, this study contributes to knowledge management theories, particularly Nonaka’s SECI model (Nonaka & Takeuchi, 1995), and provides actionable guidance for library administrators aiming to build inclusive, intelligent, and data-driven academic libraries.

Methodology

This study adopts an applied mixed-methods approach, integrating qualitative content analysis with quantitative prioritization techniques to develop a strategic framework for artificial intelligence (AI) integration in academic libraries. The research was conducted in three distinct phases to ensure systematic collection, rigorous analysis, and objective prioritization of data from strategic plans of leading university libraries. This methodology aligns with established standards in library and information science research, emphasizing reliability, transparency, and replicability (Creswell & Creswell, 2018; Bazeley, 2021).

Research Design and Data Collection

The study examined strategic plans from the top 25 universities according to the Times Higher Education (THE) 2024 World University Rankings. These institutions were selected as they represent global leaders in research, innovation, and digital initiatives, providing valuable benchmarks for best practices in academic libraries.

Data consisted of 25 strategic documents published or updated primarily between 2018 and 2024. These included standalone library strategic plans, university-level digital transformation strategies, and institutional plans with substantial library components. When a dedicated library strategic plan was unavailable, the most relevant university-wide document addressing library services or digital strategies was selected. A detailed list of all documents—including titles, publication years, document types, and access URLs (where available)—is provided in Appendix A.

Inclusion criteria were: (1) publicly available on official university or library websites; (2) written in English or with sufficient English content for analysis; and (3) explicitly addressing library services, digital transformation, scholarly communication, or emerging technologies. Documents were retrieved between October and December 2024. The final corpus totaled approximately 1,200 pages. This pragmatic approach acknowledges that not all leading universities maintain separate, recent library-specific strategic plans, but all address relevant themes at the institutional level.

Data Analysis

Phase 1: Qualitative Content Analysis Strategic documents were analyzed using conventional qualitative content analysis (Hsieh & Shannon, 2005). The process began with open coding to identify initial concepts directly from the data (e.g., “digitization of collections,” “digital preservation,” “data analytics,” “machine learning,” “AI,” “chatbot,” “algorithmic bias,” “open access”). Through axial coding, these concepts were grouped into broader categories and sub-categories. Categories were iteratively refined through constant comparison. MAXQDA 2024 software was used for systematic coding, memo-writing, and data management. Two researchers (the author and an independent coder with expertise in library science) independently coded the documents to enhance rigor. Disagreements were resolved through detailed discussion and consensus until full agreement was achieved on the final coding scheme.

Phase 2: Intercoder Reliability To assess the reliability of the coding process, a second researcher independently recoded a random sample of 20% of the documents (5 documents). Intercoder reliability was measured using Cohen’s Kappa coefficient, yielding κ = 0.737, which indicates substantial agreement (Landis & Koch, 1977). Although double-coding 20% of the corpus is a common and accepted practice in content analysis studies of complex documents, all discrepancies were resolved through consensus discussions, further strengthening the trustworthiness of the findings.

Phase 3: Quantitative Prioritization Following qualitative coding, Shannon entropy was employed to prioritize the identified components based on their frequency of mention and strategic emphasis across the 25 documents. This method was chosen because it provides an objective measure of the relative importance and distribution of priorities, accounting for both frequency and diversity rather than relying solely on raw counts. The entropy formula used was:

where represents the probability (relative frequency) of each component. Higher entropy values indicate greater diversity (less dominance) in priorities. Sensitivity analysis was then conducted by systematically removing each component one at a time and recalculating the entropy value to evaluate its impact on the overall model stability. Descriptive statistics (frequencies, percentages, mean, and standard deviation) were calculated using SPSS Version 29.

Alternative Approaches Although large language models (LLMs) could have been used for initial coding or keyword extraction, this study prioritized manual qualitative content analysis due to its superior ability to capture contextual nuances, institutional priorities, and subtle differences between “digital transformation” and explicit “AI” initiatives. The strengths and limitations of LLM-assisted approaches are discussed in the Discussion section.

Ethical Considerations

As the study utilized only publicly available documents, no institutional review board (IRB) approval was required. Institutions were not identified beyond their aggregated strategic priorities in the reporting of findings to maintain focus on patterns rather than individual cases. The research adhered to ethical guidelines for data handling and scholarly reporting (American Psychological Association, 2020).

Visualization and Rigor

To enhance methodological transparency, this research presents the three-phase research process, including data collection, coding procedures, reliability checks, and prioritization steps. This visualization supports accessibility for diverse readers and aligns with best practices in information science research.

Results

All 25 universities in the sample possessed formal strategic plans or digital transformation documents. Eighteen of these documents (72%) were published or updated after 2020, reflecting recent institutional priorities on digital futures. Plan durations varied from 4 to 10 years, with most spanning 4–6 years. The documents ranged from high-level vision statements to more detailed implementation roadmaps.

Content analysis demonstrated that all universities addressed digital transformation and emerging technologies. While the explicit use of the term “Artificial Intelligence” varied, concepts such as machine learning, data analytics, automation, and intelligent systems appeared across the sample in different forms and levels of emphasis.

Shannon entropy analysis (H = 2.259) was used to quantify the distribution of priorities. As shown in Table 2, sensitivity analysis confirmed that Digitization and Digital Preservation, along with Data Analytics, are the most influential components.

Specific AI Applications

Strategic interest in AI is growing; however, explicit implementation of advanced AI tools remains limited and in early stages for most institutions.

Ethical and Risk Considerations

Attention to ethical aspects remains emerging.

Regional Comparison

A comparative analysis of the strategic plans reveals notable regional differences in emphasis:

Note: Categorization is based on the relative emphasis and frequency observed in the reviewed strategic documents.

Interpretation of Findings

The results clearly show that leading universities are prioritizing foundational infrastructure especially digitization and long-term digital preservation as the critical first step before advancing toward AI-driven analytics and user-centered services. This phased approach provides valuable lessons for libraries in resource-limited contexts.

While strategic interest in AI is widespread, concrete deployment of advanced applications (such as chatbots or intelligent agents) is still limited. Most references to AI appear within broader digital transformation initiatives. Ethical considerations are receiving increasing attention but have not yet become a central strategic priority for the majority of institutions.

These empirical findings form the basis for the scalable strategic framework proposed in the Discussion section.

Proposed Strategic Framework for Developing Countries (e.g., Iran) Based on the observed global patterns, the following phased recommendations are suggested:

1. Phase 1: Foundation: Large-scale digitization and robust digital preservation systems.
2. Phase 2: Intelligence: Development of data analytics, AI-powered search, and user insights.
3. Phase 3: Integration & Maturity: Implementation of advanced AI tools with comprehensive ethical governance.
4. Cross-cutting Elements: Continuous staff training, international collaboration, open-source solutions, and sustainable funding models.

Discussion

This study offers a systematic analysis of strategic priorities for digital transformation and artificial intelligence integration in academic libraries through the examination of strategic plans from the top 25 universities according to the Times Higher Education (THE) 2024 World University Rankings. The findings identify six primary domains of focus: digitization and digital preservation, digital learning spaces and skills development, open access and scholarly communication, data services and analytics, AI applications and advanced technologies, and ethical and social considerations. These results provide a realistic roadmap of current institutional priorities while highlighting the substantial gap between strategic aspirations and the actual implementation of advanced AI tools.

The clear dominance of digitization and digital preservation as the foundational priority aligns strongly with previous research, which consistently emphasizes that robust digital infrastructure is a prerequisite for effective deployment of advanced AI applications (Cox et al., 2023; Huang et al., 2023). Similarly, the strong emphasis on data services, analytics, and digital learning environments reflects the growing recognition of data as a strategic asset and the need to prepare users for data-rich academic environments. However, explicit and advanced AI applications—such as chatbots, intelligent recommendation systems, or large-scale machine learning deployments—remain limited, confirming that AI adoption in academic libraries is still in an emergent phase (Hervieux & Wheatley, 2021).

Regional differences further enrich the findings. European institutions tend to prioritize infrastructure and long-term preservation, North American universities show stronger emphasis on open scholarship, innovation, and AI experimentation, while Asian leading universities place notable focus on interdisciplinary collaboration and digital skills development. This comparative perspective underscores that strategic priorities are shaped by institutional context, resources, and cultural factors.

This study contributes to the literature by moving beyond single-tool or ethics-focused analyses to provide a broad, empirical mapping of strategic priorities across leading institutions. While Cox et al. (2023) primarily addressed ethical frameworks, and Jin (2025) examined AI in knowledge management through a case study approach, the present research offers a comparative view of actual institutional plans. It complements theoretical discussions on large language models (LLMs) in libraries by demonstrating why foundational digitization continues to precede sophisticated AI implementations in practice. Unlike many previous studies that remain conceptual, this research grounds its conclusions in documented strategic intentions of elite universities.

This research advances knowledge management theory by empirically linking strategic priorities to Nonaka’s SECI model (Nonaka & Takeuchi, 1995). The strong emphasis on digitization corresponds to the Externalization phase (converting tacit knowledge into explicit digital resources), while data analytics and interdisciplinary collaboration support the Combination and Internalization phases. The proposed framework extends the SECI model by adding two critical layers: contextual adaptation for resource-limited settings and ethical governance as a cross-cutting dimension. This extension provides a more comprehensive theoretical lens for understanding AI-enabled knowledge creation and digital transformation in contemporary academic libraries.

Practical Implications

The findings offer actionable insights for academic library administrators, particularly in resource-constrained environments. The proposed scalable strategic framework adopts a phased approach that mirrors the patterns observed in leading universities:

1. Foundation Phase: Prioritize large-scale digitization and robust digital preservation systems.
2. Intelligence Phase: Build data analytics capabilities and introduce AI-powered tools for search optimization and user insights.
3. Integration and Maturity Phase: Deploy advanced applications such as chatbots and recommendation systems alongside comprehensive ethical frameworks.
4. Cross-cutting Elements: Implement continuous staff training, user feedback mechanisms, and sustainable international collaborations throughout all phases.

For developing countries such as Iran, direct transplantation of strategies from elite Western or Asian universities is neither feasible nor advisable. Significant differences in infrastructure, funding, human resources, and policy environments must be considered. Therefore, the framework stresses contextual adaptation: leveraging open-source and low-cost AI solutions, focusing on high-impact applications (e.g., AI-assisted metadata generation), and integrating training programs with national digital literacy initiatives. Ethical considerations including privacy, algorithmic bias, and transparency should be embedded from the earliest phases to build public trust and ensure regulatory compliance.

Limitations

Despite its contributions, this study has several limitations. First, reliance on publicly available strategic documents may exclude internal or unpublished plans, potentially biasing results toward more transparent institutions. Second, the sample was limited to the top 25 globally ranked universities, restricting generalizability to mid- or lower-ranked and resource-constrained institutions. Third, although intercoder reliability measures were employed, the qualitative nature of the data inevitably involves some degree of interpretive judgment. Finally, given the rapid evolution of AI technologies, strategic plans analyzed in late 2024 may already be partially outdated by the time of publication.

Conclusion

This study underscores the transformative potential of artificial intelligence in evolving academic libraries into intelligent, user-centered knowledge ecosystems. By systematically analyzing strategic plans of leading universities, it identifies digitization as the current dominant priority while revealing persistent gaps in advanced AI implementation and ethical integration. The proposed scalable and contextually adaptable framework offers a practical pathway for libraries worldwide especially in developing countries to navigate digital transformation effectively.

Theoretically, the research enriches knowledge management and information science by bridging empirical strategic priorities with established models such as Nonaka’s SECI. Practically, it equips library administrators and policymakers with realistic, phased strategies for investment and implementation. Ultimately, successful AI integration depends not only on technological capabilities but also on human capacity building, ethical responsibility, and sensitivity to local contexts.

Future research should evaluate the cost-effectiveness and real-world impact of specific AI applications, examine adoption patterns in non-elite institutions, explore hybrid LLM-qualitative analysis methods, and conduct longitudinal studies to track the evolution of library strategies over time. Bridging the gap between global best practices and local realities will be essential for creating inclusive, equitable, and intelligent academic libraries that effectively serve 21st-century scholarship and society.

Declarations

Funding: This research received no external funding.

Conflicts of Interest: The author declares no conflicts of interest.

Ethics Approval and Consent to Participate: Not applicable, as this study did not involve human or animal subjects requiring ethical approval.

Consent for Publication: Not applicable, as this manuscript does not contain any sensitive or identifiable personal data.

Availability of Data and Materials: The data supporting the findings of this study are available from the author upon reasonable request via email.

Authors’ Contributions: This research was conducted solely by the author. The author conceptualized the study, conducted the research, analyzed the data, and wrote the manuscript.