AI evidence extraction

Governing Generative Artificial Intelligence in Higher Education: A Cross-Case Thematic Analysis of Institutional Policies

Muhammad Usama Islam, Senanu Okuboyejo, Foluso Ayeni · Journal of the Association for Information Systems · 2026

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

10/10

Relevance

1/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

Methodology & findings

Study design

Qualitative cross-case thematic analysis using Braun and Clarke's (2006) six-phase thematic analysis approach.

Sample

N = 23, 1 group

Primary method

Qualitative thematic analysis following Braun and Clarke's (2006) six-phase approach. A hybrid analytical approach combining manual interpretation and AI-assisted analysis using large language models for code identification, pattern recognition, and theme development. Iterative review and cross-case validation were conducted.

Main result

The study identified five core themes characterizing institutional approaches to AI governance: "The analyzed corpus comprised 23 policy documents totaling 13,514 words, with an average length of 2,252.3 words per institution, reflecting the substantial depth and priority afforded to AI governance by these universities. The thematic analysis identified five major themes that characterize how universities conceptualize the governance of generative artificial intelligence. These themes emerged consistently across the analyzed policy documents, although the emphasis varied by institution." The findings reveal that "while all universities emphasize academic integrity, data protection, and human oversight, they differ in their approaches to governance and pedagogy. Some institutions prioritize centralized control and risk management, while others emphasize instructional flexibility and pedagogical autonomy."

Reports effect sizes.

Research paradigm

Interpretivist/Qualitative

Author conclusions

"This study examined generative artificial intelligence policies across five leading universities using thematic analysis. The findings identify five core themes that characterize institutional approaches to AI governance: institutional governance, pedagogical governance, academic integrity, data security, and human oversight. The results show that universities adopt different governance orientations while converging on key principles such as transparency, data protection, and accountability. These findings contribute to Information Systems research by conceptualizing generative artificial intelligence governance as a socio-technical phenomenon and by demonstrating the use of AI-assisted qualitative methods."

Risk of bias

Selection bias: Only five top-ranked US universities selected based on Times Higher Education World University Rankings; Geographic bias: Only US-based institutions included; Document availability bias: Limited to publicly accessible policy documents; Interpretation bias: AI-assisted analysis used for initial coding, potentially biasing theme identification toward patterns recognizable by language models; Status quo bias: Analysis limited to documented policies without examining actual implementation or user behavior; Selection bias: Only top-ranked US universities selected based on Times Higher Education rankings; Document selection bias: Only publicly available policy documents included; Limited generalizability: Sample restricted to five institutions; Analyst bias: Potential interpretation bias despite AI-assisted analysis; Selection bias: Only top-ranked US universities included based on Times Higher Education World University Rankings; Sample limitation: Analysis of only 5 institutions may not be representative of broader higher education landscape; Document selection bias: Reliance on publicly available policy documents may miss informal or unpublished governance practices; AI-assisted analysis bias: Use of large language models for coding may introduce algorithmic biases in theme identification

Limitations

"However, this study's empirical base is limited to a selective sample of five universities
Furthermore, our analysis focuses on documented policies ('policy on paper') rather than their real-world enforcement or interpretation by faculty and students ('policy in practice')
Future research should explore this distinction through longitudinal or ethnographic studies."

Open questions raised

Limited comparative analysis of AI governance across institutions
Gap between documented policies and their real-world enforcement ('policy on paper' versus 'policy in practice')
Need for longitudinal or ethnographic studies to understand implementation
Broader institutional coverage beyond five leading universities
Temporal analysis of how policies change over time as generative AI evolves
Limited research examining how universities conceptualize generative AI governance

Extracted from: pdfAgreement 65%

Explore related topics

Institutional Policy AI Governance Qualitative