Generative AI driving major shifts in teaching, research and governance worldwide

The study’s decade-long timeline shows a large dataset of 22,646 documents, but the most striking result is not the volume; it is the timing. Annual scientific production grew slowly from 2015 through 2022 and then surged sharply in 2023. The shift aligns with the global adoption of generative AI tools, which triggered new debates about ethics, pedagogy, and academic integrity.

A ten-year review of global scientific output shows an abrupt and historic rise in research on artificial intelligence in higher education beginning in 2023, marking a structural shift in how universities worldwide understand and integrate AI technologies. The findings come from the study “Mapping the AI Surge in Higher Education: A Bibliometric Study Spanning a Decade (2015–2025)”, published in Informatics, which analyzed more than 22,000 research documents to uncover patterns shaping AI’s expanding role across universities.

Using combined datasets from Web of Science and Scopus, the authors tracked publication trends, leading contributors, country-level output, top-cited work, and emerging themes. The study reveals that AI discussions in higher education remained scattered and comparatively low until 2022, followed by a dramatic acceleration in 2023 driven by the rise of generative AI systems. The evidence shows that this growth is reshaping scholarly priorities, redirecting attention toward pedagogy, academic integrity, ethics, governance, and global inequality in digital transformation.

The research identifies both promise and risk as universities face rapid disruption. It points to a widening divide between countries with strong AI capacity and those without, even as global policy frameworks call for inclusion, equity, and alignment with the Sustainable Development Goals. The map of scholarly activity shows intense clustering around a small number of nations, authors, and disciplines, indicating that AI’s influence in higher education is expanding but not evenly shared.

Accelerated growth and shifting research patterns

The study’s decade-long timeline shows a large dataset of 22,646 documents, but the most striking result is not the volume; it is the timing. Annual scientific production grew slowly from 2015 through 2022 and then surged sharply in 2023. The shift aligns with the global adoption of generative AI tools, which triggered new debates about ethics, pedagogy, and academic integrity.

The review notes that earlier analyses often portrayed AI research in higher education as steadily rising, but the data contradicts this narrative. Instead, the years before 2023 saw steady but modest activity, with limited attention from many institutions. The sudden acceleration in 2023 reflects a clear turning point as AI became embedded in teaching, learning, assessment, and research practices.

Citations per document averaged above eight across the dataset, a sign of strong uptake and cross-disciplinary engagement. The research also found more than 63,000 unique authors contributing to the field over the decade but with clear concentrations of influence. Co-authorship patterns show that while collaboration is significant, only about 12 percent of collaborations span multiple countries, leaving most work clustered within national contexts.

These findings highlight a maturing research domain where activity is increasingly global yet still fragmented by region, access, and capacity. The authors argue that the wave of new work after 2023 signals a new stage for higher education, in which AI is no longer experimental but foundational to strategic planning.

Who leads the global output and how research is distributed

The bibliometric mapping points to China and the United States as the primary engines of global research, dominating publication volume, citation impact, and international collaborations. These two countries appear consistently at the top across multiple performance metrics, reflecting their strong institutional infrastructures, AI investments, and large research ecosystems. Other contributors include the United Kingdom, Germany, Korea, Australia, Spain, Japan, Saudi Arabia, and India, though with significantly smaller outputs.

This imbalance has direct implications for equity. Regions such as Africa and South America remain underrepresented despite growing interest and demonstrated need. As the world approaches the 2030 targets of the Sustainable Development Goals, limited participation from these regions introduces critical gaps in digital readiness, education policy, and AI-driven innovation for social development. The study states that global inequalities persist despite widespread adoption of AI discourse and calls for more inclusive strategies.

On journal productivity, the study uses a performance model to identify a core group of highly active publications. Journals such as Advances in Intelligent Systems and Computing, Education and Information Technologies, Sustainability, Education Sciences, Computers and Education: Artificial Intelligence, and PLOS One emerge as central hubs for AI-related higher education scholarship. These journals form the core zone with the highest frequency of publication within the dataset.

A large portion of the most-cited work relates not only to pedagogy but also to the technical foundations of AI. Highly influential papers include landmark deep learning studies that underpin modern machine learning systems, alongside newer articles examining ChatGPT’s educational implications, academic integrity concerns, and ethical debates. Twelve of the twenty most locally cited references were published between 2023 and 2025, reinforcing the finding that the field’s intellectual center of gravity has moved rapidly toward generative AI.

The author impact analysis highlights several Asian researchers as leading contributors, reflecting both the scale and growth trajectory of AI research in the Far East. Detailed metrics show strong production and citation patterns among these authors, underscoring the region’s expanding influence in shaping global AI scholarship.

What universities are studying: Themes, ethics, and the generative AI shift

The study uses VOSviewer keyword analysis and Latent Dirichlet Allocation topic modelling to interpret thematic patterns in the literature. These analyses reveal striking consistency across the decade and across datasets.

The keyword networks show intense clustering around concepts such as artificial intelligence, students, teaching, learning, academic integrity, ChatGPT, ethics, generative artificial intelligence, and higher education. The clusters also include emerging topics like virtual reality, personalized learning, adaptive systems, gamification, blockchain applications, and data-driven instruction. Together, these patterns reflect a research agenda focused on both practical deployment and socio-ethical governance.

A more refined view comes from the LDA analysis, which detects two dominant themes:

• Pedagogical integration of generative AI tools. This theme captures learning transformation, student use, classroom tools, instructional design, and digital pedagogy. It reflects the shift toward AI-supported teaching models and the need for new frameworks to guide assessment and curriculum development.
• Ethical implications and research-focused concerns. This theme covers data governance, human oversight, academic integrity, transparency, and questions of trust. It points to a broad scholarly effort to understand how AI affects institutional responsibility, research culture, and the boundaries of academic work.

The authors argue that these themes reflect the dual challenge facing universities: adopting AI for innovation while establishing guardrails that protect fairness, accuracy, and human agency. As the literature expands, the tension between opportunity and risk becomes more pronounced.

The study asserts that universities must prepare for a deep transformation that goes beyond classroom technology. It calls for comprehensive frameworks that address governance, ethics, privacy, transparency, and long-term sustainability. It also highlights the need for ongoing involvement from policymakers, who must help steer AI adoption in ways that support national education systems and global equity.

Mapping AI’s rise in higher education is not simply a technical exercise but a significant strategic and ethical undertaking. Institutions must act as responsible stewards rather than passive adopters, ensuring that innovations serve the goals of human development, educational quality, and social justice.

Growing responsibilities and future directions

The authors note that although the research field is expanding, critical gaps remain. The dominance of a few regions means that global perspectives are still uneven. The reliance on Scopus and Web of Science also means that some regional scholarship may not yet be visible in global bibliometric maps. They argue for broader database inclusion in future work, along with more qualitative studies to understand barriers to AI integration in low-resource settings.

They call for sustained attention to ethical frameworks, curriculum reform, and capacity building as universities navigate the next wave of AI-driven change. The report stresses that future studies should monitor post-2025 trends to see whether the 2023 surge stabilizes, intensifies, or evolves into a new paradigm for higher education.

The study makes clear that AI’s rise in higher education is now irreversible. What remains uncertain is how universities will balance innovation with responsibility as they prepare students for a world increasingly shaped by intelligent systems.