How artificial intelligence is reshaping information literacy in academic libraries: A global Scientometric analysis (2020–2025)

Synthesis: Jabeen and Lux map the literature that joins AI and information literacy in academic libraries, drawing 1,669 publications from five databases covering 2020 to 2025 and analyzing them with CiteSpace and VOSviewer. The map shows a field anchored on information literacy rather than displaced by it — 811 keyword occurrences against 413 for libraries and 301 for librarians — with AI, ChatGPT and generative AI layered on as the fast-rising terms. Output is concentrated in a small set of library journals and dominated by the United States, with 179 publications against 30 for Australia, 21 for Pakistan and 20 for China. Author co-citation resolves into three discourses with no unified model, and the authors are explicit that citation visibility is not evidence of changed library practice.

Key Findings

1. The dataset moved from 6,500 identified records to 5,112 after preliminary screening and deduplication, 1,886 at eligibility and 1,669 included.
2. The Journal of Academic Librarianship is the most productive outlet with 92 publications (5.5%) and the central node of a journal co-citation network of 116 journals and 444 links.
3. Communications in Information Literacy follows with 65 (3.9%) and the Journal of Information Literacy with 56 (3.4%); College & Research Libraries has 43 (2.6%) and the Journal of Librarianship & Information Science 42 (2.5%).
4. The United States leads national output with 179 publications, ahead of Australia (30), Pakistan (21), China (20), India (19), Nigeria (16), Canada (16), South Africa (13), Bangladesh (12) and Malaysia (10).
5. South–South collaboration is visible: Nigeria–South Africa appears in 8 papers, Indonesia–Malaysia in 5 and Jordan–Malaysia in 4.
6. Information literacy remains the dominant keyword at 811 occurrences, against 413 for libraries and 301 for librarians; AI, ChatGPT and generative AI are the rising terms rather than replacements.
7. Author co-citation forms three clusters: AI literacy as measurable competence (Wang, Ng, Cetindamar), classic digital literacy (Martin, Bawden) and AI literacy in academic-library practice (Hervieux, Wheatley, Cox, with Massis also prominent) — the third and largest.
8. The most-cited recent work by time-normalized citation is instrument and framework building: Wang et al.'s AI literacy scale (TGC/t 99.33), Okunlaya et al.'s conceptual framework (93.25) and Harisanty et al.'s systematic review (74.00).
9. Keyword structure splits into an intellectual core (IL, AI, ChatGPT, chatbots, LIS, librarians, students), a practice and pedagogy cluster (academic libraries, digital literacy, curriculum, teaching methodologies, media literacy) and a methods and situations cluster (surveys, questionnaires, self-efficacy, COVID-19, online education, health literacy).
10. Ethical concerns — algorithmic bias, privacy, responsible generative-AI use, human oversight — are treated as core information-literacy content rather than an add-on.

What a citation map can and cannot show

The paper's strongest claim is structural: the field did not replace information literacy with AI literacy, it layered AI onto an existing construct. The keyword counts support that, and so does the journal concentration — the outlets publishing this work are library and information science journals, not computer-science venues. Whatever AI literacy becomes, on this evidence it is being built inside the library profession's own literature.

The weaker inference is the one readers will want to make: that libraries have changed their practice. Citation counts measure visibility, and the authors say plainly that rising attention and citation peaks do not prove extensive or substantial use of AI in library practice. The concentration in a handful of journals compounds the problem, because a field that publishes in its own venues can look active while the practice it describes remains a set of pilots.

The Global South gap is a measurement artifact and a real one

Nigeria, South Africa, Pakistan, Bangladesh and Malaysia appear in the top ten by output, and the South–South pairs are quantified. Yet the authors also note that indexing and editorial systems favor Global North journals, so participation grows faster than visibility. Both statements are true at once: the map shows a genuinely international field and understates the parts of it that publish outside the indexed venues. Reading the country ranking as a capacity ranking would repeat the error the authors warn about.

What this means for practice

• Do not read this map as an adoption survey. It measures publications and citations; a library planning a service needs its own usage data.
• Use the three clusters to place a program. Competence measurement, classic digital literacy and library practice are different conversations with different literatures; a staff development plan should say which one it is joining.
• Look outside the core journals. The concentration in a few LIS outlets means relevant work on AI in libraries is also published in education and health-information venues that this dataset only partly captures.
• Treat ethics as content, not a disclaimer. The field's keyword structure places bias, privacy and oversight inside information literacy, which is where instruction can actually address them.

Limitations

• The dataset depends on five databases and on title, abstract and keyword fields; relevant work using other terminology or indexed elsewhere is missed.
• Google Scholar and LISA were excluded, and mixing citation indexes with publisher platforms introduces coverage and publisher bias.
• Citation indicators measure visibility, not research quality or library practice; the authors state that rising attention does not prove substantial use.
• The search was executed in March 2025, so publications from the rest of that year are only partly represented.
• Keyword and co-citation maps are sensitive to threshold and clustering choices, and the three-cluster solution is one reading of the network rather than a settled structure.

Connected Concepts

• Information and Media Literacy
• Libraries
• AI Literacy
• Librarians
• Meta-Analysis and Systematic Review
• Generative AI

Citation

Jabeen, M., & Lux, C. (2026). How artificial intelligence is reshaping information literacy in academic libraries: A global Scientometric analysis (2020–2025). The Journal of Academic Librarianship, 52, 103221.