Academic Librarians' Student Engagement for Promoting Information Literacy and Artificial Intelligence Instruction in Nigerian Tertiary Institutions

Synthesis: Dada surveys 412 academic librarians across 38 Nigerian universities, polytechnics and colleges of education to ask how they engage students on information literacy and AI instruction, how competent they feel, what blocks them and what predicts effectiveness. The pattern is a competence paradox: librarians rate themselves high on traditional information-literacy delivery (M = 4.02) and database instruction (M = 4.07), but their composite AI competency sits below the scale midpoint (M = 2.86), with the weakest items being designing AI literacy objectives (M = 2.70) and assessing AI literacy outcomes (M = 2.62). Face-to-face reference and formal instruction dominate engagement; AI orientation sessions are the least-used mode (M = 2.77), and 50.7% say they rarely run them.

Key Findings

1. The survey drew 412 usable responses from 450 instruments (91.6%) across 38 federal and state institutions, using stratified random sampling by institution type, geopolitical zone and librarian cadre; fieldwork ran March–May 2023.
2. Traditional engagement dominates: in-person reference consultation has M = 4.10 (SD = 0.96) and formal library instruction M = 4.05 (SD = 0.93), against embedded librarianship M = 3.05 and AI tool orientation M = 2.77 (the lowest of eight modes).
3. For AI tool orientation sessions, 50.7% selected "rarely" and only 7.3% selected "always".
4. Self-reported competence is high for traditional work — database search and retrieval instruction M = 4.07 (61.2% high) and information-literacy instruction M = 4.02 (58.7% high).
5. Composite AI competency is below the midpoint at M = 2.86 (SD = 0.71); understanding AI concepts M = 2.88, evaluating the credibility of AI-generated content M = 3.07, teaching ethical AI use M = 3.00, integrating AI into information-literacy sessions M = 2.78.
6. The weakest items are instructional design and assessment: designing AI literacy learning objectives M = 2.70 (47.1% low) and assessing students' AI literacy outcomes M = 2.62 (51.9% low).
7. Infrastructure is the top barrier: inadequate internet connectivity and digital infrastructure drew 87.4% agreement (M = 4.38, SD = 0.74), followed by the absence of a formal AI literacy curriculum in LIS training at 84.2% (M = 4.27) and limited professional development at 82.5% (M = 4.20).
8. Attitudinal barriers rank last: student low awareness or motivation 71.4% (M = 3.89), heavy workload 68.9% (M = 3.79) and ethical or copyright concerns 61.2% (M = 3.55).
9. The regression model explains 52% of variance in instructional effectiveness (R² = 0.52, F(3, 408) = 148.27, p < 0.001): engagement frequency β = 0.41, prior AI training β = 0.33 and perceived administrative support β = 0.29, all significant.
10. The author recommends embedding AI literacy in library frameworks, mandating structured professional development and adding AI instruction to national library standards.

Confidence and competence diverge

The most useful result is the gap between two self-ratings on the same instrument. Librarians who feel competent teaching database search do not feel competent designing AI literacy instruction or assessing whether it worked — and those are not the same skill at a higher difficulty. One is a delivery skill, the other a design skill, and the survey locates the deficit precisely at the design and assessment end.

That matters because the barriers list is often read as a funding story. Infrastructure is indeed the top-rated barrier, but the next two are curricular and developmental: no formal AI content in LIS training and limited professional development. A library that solves connectivity and buys tools without touching those will move the infrastructure score and leave the competency score where it is.

What the regression does and does not establish

Engagement frequency is the strongest predictor of self-reported effectiveness, followed by prior training and administrative support. All three are perceptions measured in the same instrument at the same moment, which means the model describes a coherent attitude structure rather than a causal chain. A librarian who runs more sessions and has had training reports that the sessions work; whether students learned more is not measured.

The study is also silent on sampling and response bias: a 91.6% response rate is presented as a strength, and no limitations section discusses generalizability. The population is one country's academic librarians, surveyed in 2023 before the current generation of agentic tools. Those boundaries should travel with every number.

What this means for practice

• Train for design and assessment, not awareness. The weakest self-rated skills are writing AI literacy objectives and assessing outcomes; a workshop on tools does not address either.
• Make AI orientation a scheduled mode. It is the least-used engagement mode in the survey, which suggests it is nobody's assigned job rather than that it was tried and rejected.
• Fix infrastructure and curriculum together. Connectivity is the top barrier, but the curricular gap ranks second; treating either alone leaves the other in place.
• Give the work administrative cover. Perceived support is a significant predictor; a librarian running AI instruction without a policy or a mandate is carrying a risk the institution has not accepted.

Limitations

• The design is a single cross-sectional survey from March–May 2023; it cannot support causal claims, and the author describes the work as a diagnostic snapshot.
• All competency and effectiveness measures are self-report, and the study does not validate them against observed teaching or student outcomes.
• The article has no dedicated limitations section and does not discuss sampling, response bias or generalizability.
• The reported item counts in the method (12 engagement items, 14 barrier items) do not match the eight modes and ten barriers reported in the tables.
• The sample is confined to Nigerian tertiary institutions and predates the current generation of generative and agentic tools.

Connected Concepts

• Libraries
• Information and Media Literacy
• Librarians
• AI Literacy
• Higher Education
• AI Governance

Citation

Dada, K. S. J. (2026). Academic Librarians' Student Engagement for Promoting Information Literacy and Artificial Intelligence Instruction in Nigerian Tertiary Institutions. International Journal of Educational Qualitative Quantitative Research, 5(1), 105–118.