AI governance readiness, not just technology, linked to student retention intentions in Kazakhstan

A new study from Kazakhstan suggests that keeping university students enrolled may depend not on the artificial intelligence tools themselves, but on the institutional machinery that governs them. Researchers at Kazakh-British Technical University in Almaty surveyed 897 students, faculty members, and administrators across public and private higher education institutions and found that what they call AI Governance Readiness — an institution’s capacity to establish, coordinate, implement, and monitor responsible AI governance arrangements — was strongly associated with both AI-enabled administrative efficiency and the AI fluency of the people who use these systems. Those two capabilities, in turn, were linked to students’ intentions to stay at their institutions. The findings, published in Frontiers in Education, reframe AI governance as a form of institutional capital rather than a mere compliance exercise.

The research team, led by Assel Kenesovna Jumasseitova, Rajasekhara Mouly Potluri, and Maira Khussainova, grounded their work in two established theories of organizational strategy: the Resource-Based View, which treats valuable, organizationally embedded resources as sources of institutional advantage, and Dynamic Capabilities Theory, which explains how organizations transform, integrate, and reconfigure resources in response to changing technological conditions. Within this framework, AI Governance Readiness represents the institutional capacity, while Administrative Efficiency through AI and AI Fluency Orientation represent the operational and human capabilities through which that capacity is expressed. Institutional Theory serves as a complementary contextual lens, explaining why universities develop such governance capacity in response to regulatory, normative, technological, and legitimacy pressures, though it is not treated as an additional causal mechanism in the structural model.

The distinction between AI governance and AI governance readiness is central to the study. AI governance refers to the operating framework of policies, processes, risks, accountability mechanisms, ethical safeguards, and oversight arrangements through which AI is managed. AI Governance Readiness, by contrast, refers to an institution’s capacity and preparedness to establish and operationalize that framework. The authors argue that this capacity can be translated into human and administrative capabilities — and that the value of AI in higher education depends on how efficiently institutions manage and integrate these interdependent factors, not on technology adoption alone. AI technologies, in their account, are resources that require governance; governance is the system through which those resources are managed responsibly.

To test their model, the researchers used a quantitative, cross-sectional survey design. The questionnaire, originally written in English and translated into Russian and Kazakh with back-translation to ensure cultural and linguistic fit, measured five constructs: AI Governance Readiness, Administrative Efficiency through AI, AI Fluency Orientation, Student Retention Intentions, and Equity Perception. Respondents rated their agreement with statements on five-point Likert scales, and validation items on institutional AI practices allowed triangulation with reported perceptions. Data were collected through a multi-channel online approach using institutional email lists, academic and student networks, social media, and QR code access points. After rigorous screening, cleaning, and validation — including the rejection of 42 responses — 897 responses were retained. The study received ethics approval from the Business School Ethics Committee at Kazakh-British Technical University, and participation was voluntary, anonymous, and informed.

The analysis employed Partial Least Squares Structural Equation Modeling, a statistical technique well suited to models with multiple mediation and moderation pathways, using 5,000 bootstrap resamples to assess significance. The measurement model performed well against standard criteria: Cronbach’s alpha reached 0.84, McDonald’s omega 0.88, and composite reliability 0.88, all comfortably above the conventional 0.70 threshold. Average variance extracted was 0.57, exceeding the 0.50 benchmark for convergent validity, and all Heterotrait-Monotrait ratios fell below 0.85, confirming that the constructs were empirically distinct — although the authors note that the AI Fluency Orientation–Equity Perception value of 0.8496 sits very close to the threshold and should be interpreted cautiously. Variance inflation factors ranged from 1.36 to 2.84, well below the 3.3 cutoff, indicating no critical collinearity.

The structural results substantiated six of the seven proposed relationships. AI Governance Readiness was positively associated with Administrative Efficiency through AI (β = 0.591, p < 0.001) and with AI Fluency Orientation (β = 0.699, p < 0.001). Both capabilities were, in turn, positively associated with Student Retention Intentions: AI Fluency Orientation with a coefficient of 0.367 and Administrative Efficiency through AI with 0.404, both significant at p < 0.001. The model explained 48.3 percent of the variance in retention intentions, 48.9 percent in AI fluency orientation, and 35.0 percent in administrative efficiency — moderate to substantial explanatory power by behavioral research standards. Supplementary fit indices supported the model: the standardized root mean square residual was 0.067, below the 0.08 reference value, the Normal Fit Index reached 0.91, and all Stone-Geisser Q² values were positive, confirming predictive relevance, with effect sizes ranging from 0.05 to 0.96.

The mediation analyses proved particularly revealing. AI Governance Readiness was indirectly associated with Student Retention Intentions through AI Fluency Orientation, with an indirect effect of β = 0.433, and through Administrative Efficiency through AI, with an indirect effect of β = 0.375, both statistically significant. In other words, the link between governance readiness and students’ intention to stay appears to operate largely through the development of human competencies and streamlined, AI-enabled administrative processes — such as predictive analytics and early warning systems that can identify at-risk students and trigger timely, targeted support — rather than through governance arrangements alone. This capability-orchestration pattern aligns with prior research showing that technological investments translate into better outcomes when paired with digital competencies and efficient processes.

One hypothesized relationship did not hold. Equity Perception, which captures perceptions of the fairness, inclusion, and transparency of AI systems, did not significantly moderate the relationship between Administrative Efficiency through AI and Student Retention Intentions (β = −0.011, p = 0.444). The authors offer a provisional explanation: equity may not yet be salient enough in respondents’ experiences of AI-mediated institutional processes. They stress that this explanation is tentative and that the finding does not diminish the normative importance of equity, which they argue should be embedded proactively within governance arrangements — encompassing procedural fairness, transparency of AI-mediated processes, student data protection, and opportunities for human review or appeal — rather than treated as a conditional consideration that automatically shapes outcomes.

The study’s practical implications point toward coordinated institutional investment. Universities, the authors suggest, should establish clear AI policies, ethical guidelines, accountability mechanisms, and human oversight as AI becomes embedded in student-facing processes, while strengthening AI fluency through targeted training for academic and administrative staff and appropriate support for students. Interoperable AI systems, data integration, and responsive student-support processes can reinforce administrative efficiency. Policymakers, meanwhile, can complement institutional efforts with regulatory frameworks and national strategies promoting responsible, transparent, equitable, and privacy-conscious AI use in higher education. The authors also emphasize that AI-enabled university systems should complement rather than replace human support, and that responsible institutional governance can build broader societal trust in AI.

The authors are careful to delineate the study’s limitations. The cross-sectional design precludes causal inference, and the non-probability purposive sampling design — with 99.22 percent of respondents from institutions in Almaty, 80.38 percent under age 20, 85.94 percent undergraduates, and heavier participation from digitally advanced institutions — limits national representativeness and generalizability. Self-reported data may carry bias, which structural equation modeling only partially addresses, and the model omits factors such as pedagogical innovation, leadership, and student psychological engagement. Future research, the team recommends, should adopt longitudinal and panel designs, incorporate objective institutional indicators, test the model across multiple countries and cultural settings, and extend it with additional mediating and moderating factors such as trust in digital environments, institutional culture, and attitudes toward AI ethics. For now, the study offers emerging higher education systems a governance-oriented perspective: AI’s educational value, in this account, is orchestrated through institutional capability rather than purchased off the shelf.

Subject of Research: AI governance readiness as institutional capital and its association with student retention intentions in Kazakhstan's higher education

Article Title: AI governance readiness as institutional capital: a capability-orchestration framework for student retention intentions in Kazakhstan's higher education

Article References: Jumasseitova, A. K., Potluri, R. M., & Khussainova, M. (2026). AI governance readiness as institutional capital: a capability-orchestration framework for student retention intentions in Kazakhstan's higher education. Frontiers in Education, 11, Article 2002812. https://doi.org/10.3389/feduc.2026.2002812

Image Credits: AI Generated

DOI: 10.3389/feduc.2026.2002812

Keywords: AI governance readiness, higher education, student retention, Kazakhstan, institutional capital, AI fluency, administrative efficiency, PLS-SEM, equity perception, dynamic capabilities, resource-based view, cross-sectional survey