CoChat

Six Institutional Intervention Areas to Support Ethical and Effective Student Use of Generative AI in Higher Education: A Narrative Review

Open access

The DOI matches its Crossref record; 153 of the 153 references it cites were found in a registry or an index; 51 of the 116 sources its claims cite were checked for support.

Abstract

The integration of generative AI tools, such as ChatGPT, Gemini, and DeepSeek, into higher education offers transformative opportunities for personalised learning and academic productivity. However, their unregulated use raises concerns about academic integrity, critical thinking, and educational equity. This systematic review synthesises insights from 96 peer-reviewed articles, identifying six key intervention themes, namely, curriculum integration, policy and governance, faculty development, student-centred strategies, assessment adaptation, and technological infrastructure. Together, these themes form a comprehensive intervention framework designed to guide students’ ethical and effective engagement with AI. This review highlights the need for institutions to move beyond fragmented policies, fostering systemic cultural and pedagogical change to align AI use with authentic learning outcomes. By bridging theoretical gaps and providing actionable strategies, this framework equips educators and policymakers to scaffold responsible AI integration across diverse higher education contexts.

Overview

CoChat Summary

TL;DR

This narrative/systematic review synthesises 96 peer-reviewed articles (2023–2025) to propose a consolidated framework of six institutional intervention areas—curriculum integration, policy and governance, faculty development, student-centred strategies, assessment adaptation, and technological infrastructure—for guiding ethical and effective student use of generative AI in higher education. The authors argue institutions must move beyond fragmented, compliance-oriented policies toward systemic, pedagogy-led cultural change.

Key findings

  • The review identified six overarching intervention themes, decomposed into 38 codes and 18 sub-themes, forming a single consolidated framework (Table 2).
  • Of the 96 included articles, 8 were published in 2023, 41 in 2024, and 47 in 2025; 78 were empirical, 12 conceptual, and 6 systematic reviews. Among empirical studies, mixed-methods was most common (n = 34), followed by quantitative (n = 28) and qualitative (n = 16).
  • Articles originated from institutions across 50 countries, most frequently the United States (n = 16), Australia (n = 9), China (n = 9), and Saudi Arabia (n = 8), followed by the UK (n = 7) and UAE (n = 5). Empirical data collection was most represented by China, Saudi Arabia, and the USA (n = 8 each), then the UK (n = 6), Hong Kong (n = 5), and Australia, India, and the Philippines (n = 4 each).
  • The framework's themes span sub-themes such as embedding AI literacy, ULO/CLO alignment, institutional AI policies, transparent communication, stakeholder involvement, professional training, resource provision, ethical AI use education, skill development, reflective practices, AI-resilient assessment design, feedback and monitoring, alternative evaluation methods, tool accessibility, LMS integration, and data privacy and security.

Methods

  • Six-phase systematic literature review: scoping review, comprehensive search, quality assessment, data extraction, data synthesis, and write-up.
  • Searches in SCOPUS and Web of Science using the term "ethical use of AI in education"; SCOPUS yielded 3662 initial results narrowed to 57 included articles, and Web of Science yielded 1959 narrowed to 39 (after removing 24 duplicates), for 96 total. Inclusion was restricted to English-language, Q1-ranked (SJR) peer-reviewed journal articles published 2023–2025 addressing university students and student-focused or pedagogical interventions.
  • Data were extracted and coded using iterative thematic analysis (Braun & Clarke, 2006; Cong-Lem et al., 2025), with three authors independently coding an initial subset, then the remainder, with discrepancies resolved by discussion; themes were validated by two senior academics.

Limitations

  • Stated by the authors: reliance solely on Q1-ranked SCOPUS/Web of Science studies may exclude grey literature, emerging contexts, and alternative epistemologies; the synthesis does not test intervention effectiveness in real settings; no discipline-specific application models are provided; the framework is descriptive and does not specify causal or directional relationships among themes; rapid AI evolution may render recommendations temporally limited; and high-income contexts dominate researcher affiliations and data collection sites (approximately 45–50% of empirical contributions), with no studies involving data collection in low-income countries.
  • Evident in the text: the paper is titled a "narrative review" but describes itself as a systematic review, and the introduction refers to "99 peer-reviewed articles" while the abstract, findings, and conclusions report 96—an internal inconsistency in the reported sample size.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3390/educsci16010137

Cites

153 of 153 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

32 Valid · 11 Not valid · 16 Not checked · 95 Own results and interpretations

35 more claims · 15 with their cited sources checked

Create a free account to see all 154 claims and every source verdict.

See all claims — free account
  • Probably not a substantial claim. 91
  • It may state two separate claims as one. 45

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Show all 153 references

Research works printed but not linked: 4

  1. Bandura, A., and Walters, R. H. (1971). Social learning theory, Prentice Hall.
  2. Chu, 2025, How GenAI-supported multi-modal presentations benefit students with different motivation levels: Evidence from digital storytelling performance, critical thinking awareness, and learning attitude, Educational Technology & Society, 28, 250
  3. Lin, X., Chan, Y., Sharma, S., and Bista, K. (2024). ChatGPT and global higher education, STAR Scholars Press.
  4. Rogers, E. M. (1995). Diffusion of innovations, Free Press.

Cited by

See all 23

Metadata

Authors
Jayasinghe, S., Gamage, K. A. A., Yang, D., Cheng, C., Disanayake, C., & Apeji, U. D.
Affiliations
  • Southampton Solent University (GB)
  • University of Glasgow (GB)
  • Yangzhou University (CN)
  • NSBM Green University (LK)
Corresponding author
Gamage, Kelum A. A.
Venue
Education Sciences
Publisher
MDPI AG
Published
2026-01-16
Volume
16
Issue
1
Pages
137
ISSN
2227-7102 (ISSN-L)
OpenAlex
W7124438828
Cited by
23
References
118
Citations by year
2026: 23
FWCI
39.78
Type
journal article
Language
en
Open access
Locations
  • Education Sciences (publishedVersion, cc-by)
  • ENLIGHTEN (Jurnal Bimbingan dan Konseling Islam) (acceptedVersion, cc-by-sa)
  • Solent University Research Portal (Solent University) (acceptedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Social Sciences › Education › Artificial Intelligence in Education
  • Social Sciences › Education › Academic integrity and plagiarism
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
Keywords
generative AI in higher education, AI ethics, academic integrity, ChatGPT, faculty development, curriculum integration, assessment adaptation, student-centered learning, educational equity, critical thinking

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-06, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-06 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.