AI evidence extraction

Governing generative AI in higher education: a global Delphi study on policy and practice

Helen Crompton, Diane Burke, Christine Nickel, Aras Bozkurt, Fengchun Miao, Mike Sharples et al. · International Journal of Educational Technology in Higher Education · 2026

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

9/10

Relevance

0/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1186/s41239-026-00602-z

Methodology & findings

Study design

Three-phase Delphi study combined with grounded coding and collective writing.

Main result

The study found that "the majority of the panelists (88%) indicated that effective GenAI implementation in HE requires a combination of formal policies and flexible guidelines, rather than prioritizing one form of governance." Additionally, the research identified eight core thematic areas for GenAI governance: "academic integrity, ethical and responsible use, privacy and data protection, equitable access, GenAI literacy, integration strategy, human oversight and accountability, and institutional support and infrastructure."

Research paradigm

Interpretivist/constructivist

Author conclusions

"For GenAI to continue to transform HE, institutions will need to adopt flexible, inclusive, and evidence-based approaches that respond to the shifting technological and pedagogical terrain." The authors conclude that "effective policy must balance innovation with academic integrity, student agency with equitable access, and technological opportunity with institutional accountability." They emphasize that "The findings of this study reveal that effective policy must balance innovation with academic integrity, student agency with equitable access, and technological opportunity with institutional accountability" and stress that institutions should "actively integrate the voices of the student body, as their lived experiences with GenAI are essential for assessing the ecological validity of these frameworks."

Risk of bias

Selection bias: Purposive sampling of experts may not represent all perspectives in higher education; Participant pool limited to HE technology experts; excluded students, policymakers, employers, and technology developers; Attrition potential: study required participation across all three phases (85% completion rate); Temporal bias: findings represent expert perspectives at a specific point in time during rapid GenAI development; Geographic bias: while claiming international diversity, representation from low-income contexts limited to 3 panelists affiliated with multilateral organizations; Selection bias: Limited to HE experts only; excluded perspectives of students, policymakers, employers, and technology developers; Potential homogeneity bias: Panel comprised primarily of academics (24/35) and organization leaders affiliated with established institutions; Geographic representation: Although diverse across 22 countries and 6 continents, representation may not reflect all global contexts equally; Temporal validity: Findings captured at a specific point in time during rapidly evolving GenAI landscape; Selection bias: Purposive sampling of experts may not represent full range of HE perspectives; Attrition not reported: Study states 35/41 (85%) participated but does not analyze dropout characteristics; Expert consensus bias: Panel composed exclusively of academics and organizational leaders, excluding student and employer perspectives; Geographic representation may not reflect adequate coverage of low-income contexts (only 3 panelists from multilateral organizations serving low-income countries); Potential self-selection bias in willing participants

Limitations

"This study defines the core competencies for GenAI in HE
Intentionally, it does not list all the specific policies and guidelines in detail, as those details would rapidly become irrelevant and/or obsolete." Additionally, "While the Delphi method and collective writing approach ensured methodological rigor and inclusivity of expert perspectives, the participant pool, although geographically diverse, was limited to HE experts
The perspectives of other crucial stakeholders, such as students, policymakers, employers, and technology developers, were not directly represented." The authors also note that "The rapid pace of GenAI innovation means that any policy or guideline recommendations risk becoming outdated as technologies, capabilities, and ethical implications continue to evolve."

Open questions raised

Lack of practical implementation studies examining how GenAI governance operates across diverse HE contexts
Need for longitudinal research tracing how institutions adapt policies over time as technologies and ethical norms evolve
Absence of comparative studies across regions clarifying cultural, regulatory, and infrastructural differences in GenAI governance
Limited equity-focused research assessing whether policies bridge or widen educational divides
Missing student and community voice perspectives on GenAI governance implementation
Need for systems-level research informed by frameworks such as SETI model

Extracted from: pdfAgreement 80%

Explore related topics

Institutional Policy AI Governance Qualitative