Mapping the Integration of AI into Business Education: Insights from a Decade of Research

Synthesis: Espino & Espino (2026), Journal of University Teaching and Learning Practice 23(6). A bibliometric systematic review of 213 peer-reviewed articles (2015–2024) drawn from the Dimensions database, analyzed with bibliographic coupling and co-word analysis in VOSviewer. The review maps AI in education as applied specifically to business education, surfacing four major research clusters — (1) AI-driven business-education transformation, (2) innovative digital pedagogies, (3) AI-enhanced personalization of learning, and (4) business education aligned with the digital economy — plus three co-word trends (technological transformation, integration of generative AI tools, and advancing educational quality). Persistent gaps across clusters center on curriculum coherence, educator readiness, and Assessment validity. The authors reframe AI integration as a structural pedagogical reconfiguration rather than tool adoption.

Key Findings

• Four dominant research clusters emerge from bibliographic coupling. Cluster 1 (business education transformation in the AI era) foregrounds AI's role in curriculum and knowledge construction; Cluster 2 (innovative digital pedagogies) ties AI to active and experiential learning approaches like Simulation; Cluster 3 (AI-driven personalization) centers on personalized learning via chatbots, avatars, and learning analytics; Cluster 4 (transformation for the digital economy) links AI to entrepreneurship education and workforce readiness. Together they show research shifting from discrete tool experiments to systemic, pedagogically grounded concerns.

• Co-word analysis reveals three conceptual trends. From 6,731 extracted keywords (65 meeting a ≥10-occurrence threshold), three clusters form: (1) technological transformation in business education, (2) curriculum integration of generative AI tools (ChatGPT, chatbots), and (3) advancing quality in business education through AI. The field's most co-occurring terms — artificial intelligence (119), education (110), student (106) — underscore both its technological foundations and its learner-centered orientation.

• AI is reframed as a pedagogical reconfiguration, not a technical add-on. Across both analyses, AI increasingly functions as a structural element of educational environments — reshaping how knowledge is constructed, demonstrated, and evaluated — rather than an auxiliary enhancement. This signals a maturation from tool-centric inquiry toward capability-centered inquiry.

• Persistent gaps recur across all clusters. Three interlocking challenges dominate: curriculum coherence (AI tools adopted in isolation rather than aligned with outcomes and Assessment frameworks), educator readiness (disparities in pedagogical competence, ethical judgment, and Assessment literacy), and Assessment validity / instructional authenticity (harder to evidence genuine learning and maintain academic integrity in AI-mediated environments).

• Generative AI demands authentic, process-oriented assessment. The co-occurrence of generative-AI themes with Assessment and integrity terminology signals an emerging tension: instead of misconduct prevention, the literature increasingly stresses authentic and process-visible assessment aligned with intended learning outcomes — tasks requiring interpretation, contextual application, and iterative refinement that resist automation.

• Bibliometric synthesis complements narrative reviews. Because the dataset is large (213 articles), longitudinal (2015–2024), and methodologically replicable, the review offers a system-level map that narrative literature reviews miss, providing an empirical foundation for future research agendas and strategic educational reform.

Study Design & Method

• Data source & scope. Peer-reviewed journal articles from the Dimensions database (selected for its open-access model and multidisciplinary coverage), restricted to 2015–2024 for temporal consistency and scholarly rigor; book chapters and conference proceedings excluded. A refined Boolean search string combined AI terms (e.g., "artificial intelligence," "machine learning," "ChatGPT," "generative AI") with business-education terms (e.g., "business education," "entrepreneurship education," "business curriculum").

• Bibliographic coupling. Used to assess thematic similarity between documents based on shared citations; after thresholds, 33 highly interconnected publications were retained, clustering into four groups. Top coupling strength came from Somià & Vecchiarini (2024) and Vecchiarini & Somià (2023).

• Co-word analysis. Applied to terms from titles/abstracts to uncover conceptual structure and emerging research fronts; binary counting minimized bias from longer abstracts, and a thesaurus cleaned irrelevant terms, yielding three clusters.

• Tools. All network visualizations built in VOSviewer, enabling Visualization of research clusters, intellectual structure, and cross-disciplinary intersections.

What this means for practice

• Curriculum designers. Align each AI tool with published program outcomes and Assessment criteria before adoption, and prefer process-visible tasks over artifact-based ones, since the review's most persistent failure is isolated tool use inside otherwise unchanged curriculum and assessment structures.
• Administrators. Fund sustained educator development in pedagogical competence, ethical judgment, and assessment literacy ahead of tool procurement, because educator readiness recurs as an unresolved gap in every cluster rather than a one-off training need.
• Researchers. Build next studies on the four bibliographic-coupling clusters and three co-word trends, targeting the curriculum-coherence, educator-readiness, and assessment-validity gaps instead of adding another single-tool case.

Limitations

• The map rests on journal articles indexed in one database (Dimensions, 2015–2024), with book chapters and conference proceedings excluded, so it reflects journal publishing rather than the full literature on the topic.
• Thresholding narrowed the coupling analysis from 213 retrieved documents to 58 meeting the cited-reference criterion and then 33 retained for clustering, so the four clusters summarize a subset of the field.
• Bibliographic coupling and co-word analysis describe citation and keyword structure only; they carry no information about study quality, effect sizes, or actual learning outcomes.

Citation

Espino, L. C., & Espino, C. L. (2026). Mapping the integration of AI into business education: Insights from a decade of research . Journal of University Teaching and Learning Practice, 23(6).