AI evidence extraction

AI Governance in Higher Education: Linking Policy,Cognitive Effort, and Academic Confidence

Elizabeth Mehl, Pitso Tsibolane · Journal of the Association for Information Systems · 2025

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

8/10

Relevance

1/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

Methodology & findings

Study design

PRISMA-based systematic review of 73 empirical studies examining the relationship between institutional AI policies and student learning outcomes, utilizing theoretical frameworks from Cognitive Load Theory, Bloom's Taxonomy, and self-efficacy theory to synthesize findings across identified policy types..

Sample

N = 73, 1 group

Primary method

PRISMA-based systematic review methodology. Specific statistical analysis methods are not detailed in the abstract provided.

Main result

The study identified three policy types with distinct outcomes: "restrictive policies limit digital literacy, permissive policies foster over-reliance, and guided frameworks support higher-order engagement and calibrated confidence." These findings were derived from examining 73 studies that investigated how institutional AI policies shape cognitive effort and academic confidence in higher education contexts.

Reports effect sizes.

Research paradigm

Positivist/empiricist with theoretical framework integration

Author conclusions

The authors conclude that the research "contributes by operationalising cognitive effort and confidence as constructs for empirical testing, extending IS debates on AI governance, and offering evidence-based recommendations for AI-resilient assessment and institutional policy design." This framing positions the work as advancing both theoretical understanding and practical institutional guidance for managing generative AI in higher education.

Risk of bias

Publication bias (systematic reviews typically over-represent published studies); Selection bias in study inclusion criteria; Heterogeneity in outcome measurement across included studies; Potential reporting bias in how AI policies and outcomes were documented; Not explicitly stated in the abstract. Potential risks inherent to systematic reviews include publication bias, selection bias in study inclusion, and heterogeneity in study designs across the 73 included studies.; Publication bias (systematic reviews typically include published studies only); Heterogeneity across reviewed studies in design and measurement; Potential reporting bias in included studies

Open questions raised

The study identifies the need for empirical testing of six proposed hypotheses, development of validated measures for cognitive effort and academic confidence in AI contexts, and investigation of demographic and literacy moderators in policy-to-outcome pathways.
The paper identifies that effects of institutional AI policies on student learning remain unclear. It highlights the need for empirical testing of the proposed six testable hypotheses and evidence-based policy design for AI-resilient assessment in higher education.
The paper identifies gaps by proposing "six testable hypotheses" and highlighting "demographic and literacy moderators" as areas requiring empirical investigation. The authors note that "their effects on student learning remain unclear," indicating a gap in current understanding of how AI governance frameworks impact student outcomes.

Data: not_statedCode: not_statedExtracted from: pdfAgreement 55%

Explore related topics

Institutional Policy AI Governance Meta-Analysis