Quality Assurance 5.0 Links AI to Continuous University Improvement

A conceptual study proposes that universities could move beyond periodic, document-driven quality checks by combining artificial intelligence with continuous institutional learning. The proposed “Quality Assurance 5.0” framework treats AI as a system for generating timely evidence, rather than as an autonomous decision-maker, and relies on learning-oriented cultures, strategic leadership and evidence-informed governance to turn that evidence into action. The paper presents the framework and its relationships as testable propositions; it does not report an implemented system or demonstrate improved university performance.

The model was developed through a purposive literature search across Scopus, Web of Science, Google Scholar and ERIC, supplemented by publications from international and national quality-assurance bodies. Searchers prioritized peer-reviewed and institutional work published from 2023 to 2026, while retaining foundational literature on organisational learning, continuous improvement and total quality management. Of 612 records initially returned, the authors removed 134 duplicates, screened 478 titles and abstracts, examined 96 full texts and retained 61 sources, plus nine foundational works identified through citation chaining. Two rounds of independent coding then mapped the literature to four proposed elements: AI-enabled quality systems, continuous institutional learning, evidence-informed governance, and strategic leadership and learning culture.

At the center of the framework is a feedback loop. Predictive models, real-time monitoring dashboards and automated feedback systems are proposed to generate continuous evidence about academic and administrative quality. Organisational learning then provides the route by which staff interpret, share and institutionalise that evidence, while formal governance structures turn it into decisions, policies and accountability measures. Strategic leadership and a learning culture are proposed to influence how strongly the system works. The authors argue that AI-generated insights should not be expected to improve an institution directly: their proposed model instead places institutional learning between technological capability and outcomes such as agility, innovation, accountability and resilience.

The framework draws on three established traditions. Organisational Learning Theory explains how knowledge moves from individual observation through group interpretation to organisation-wide practice. Total Quality Management contributes principles including senior-management commitment, employee involvement, stakeholder inclusion and shared responsibility for quality. Continuous Improvement Theory supplies the Plan-Do-Check-Act cycle, in which planned interventions are implemented, evaluated and adjusted. In the proposed model, AI could compress the intervals at which the “Check” stage occurs, allowing quality review to operate continuously rather than only during annual exercises or multi-year accreditation cycles. The paper describes this as a socio-technical synthesis: software can accelerate evidence collection, but people and institutions must interpret and act on it.

Four propositions define the proposed relationships. First, AI-enabled systems are expected to improve the timeliness and granularity of evidence available for quality decisions, without that improvement automatically translating into better institutional outcomes. Second, continuous institutional learning is proposed to mediate any effect of AI on institutional agility, innovation, accountability and resilience. Third, strategic leadership and learning culture are expected to moderate that mediated relationship, so identical systems and routines may produce different results depending on management commitment, psychological safety and stakeholder inclusion. Fourth, evidence-informed governance is given a recursive role: decisions alter the data subsequently collected and the routines used to interpret future evidence, creating a continuous improvement cycle.

Technically, the proposed AI pillar would connect data from student-information, learning-management, financial and human-resource systems, potentially identifying changes in engagement, complaints or module performance before they become visible in annual surveys or accreditation self-studies. Predictive models could estimate attrition risk, projected completion rates or employability trajectories while a programme is running, allowing earlier intervention. Automated feedback systems could provide students with more timely responses and reveal recurring error patterns across a cohort for curriculum review. National frameworks already described in the literature, including one from Qatar’s qualifications and accreditation body, illustrate how AI can support accreditation reporting and monitoring while retaining human responsibility for substantive academic judgment.

Several risks accompany that technological promise. Educational algorithms can reproduce or amplify inequalities when they use historical institutional data, easily available demographic variables or outcome categories that do not capture a student’s actual circumstances. One fairness metric may also fail to transfer reliably between institutions because power relations and data conditions differ. Reviews cited by the paper find that benefits from AI-enabled educational systems coexist with infrastructure constraints, staff resistance, privacy concerns and weak change-management capacity. The authors therefore specify seven safeguards: assigned responsibility for algorithm-supported decisions; human review and appeal; documented data provenance; restricted access to individual-level predictions; independent subgroup audits; stakeholder participation; and transparent procedures for suspending harmful systems.

The framework could be adopted in stages, especially where digital infrastructure or funding is limited. Universities might first strengthen cross-unit review forums, knowledge repositories, feedback routines and formal Plan-Do-Check-Act processes, then add predictive capabilities as conditions permit. The paper argues that organisational learning may be a comparatively accessible starting point because staff can improve knowledge-sharing practices without purchasing sophisticated analytics. Accreditation agencies and regulators could also consider whether continuous monitoring and documented learning capacity should complement periodic self-study, while maintaining protections for human academic judgment. The proposed model applies to academic programmes, administration, student support, assessment and accreditation, but no particular institutional deployment was evaluated.

The principal limitation is that QA 5.0 remains a literature-grounded model whose four propositions await empirical testing. The authors call for validated measurement instruments, longitudinal comparisons and structural-equation studies across universities with different sizes, resources and regulatory settings, including cases in Sub-Saharan Africa. They also note that much of the evidence about algorithmic bias, resistance and implementation failure comes from Global North institutions, leaving its transferability elsewhere uncertain. The paper’s strongest claim is therefore not that AI has already transformed university quality assurance, but that technology and organisational learning should be designed as reciprocal components of adaptive governance. Whether that combination produces measurable, equitable and sustainable gains is an open question for future research.

Subject of Research: A conceptual framework for integrating artificial intelligence and continuous organisational learning into higher-education quality assurance.

Article Title: Quality assurance 5.0: reframing quality management in higher education through artificial intelligence and continuous institutional learning

Article References: Swem, T., & Gasa, V. (2026). Quality assurance 5.0: reframing quality management in higher education through artificial intelligence and continuous institutional learning. Frontiers in Education, 11, Article 1962474. https://doi.org/10.3389/feduc.2026.1962474

Image Credits: AI Generated

DOI: 10.3389/feduc.2026.1962474

Keywords: higher education, quality assurance, artificial intelligence, institutional learning, predictive analytics, quality management, university governance, algorithmic bias, continuous improvement, Quality, assurance, reframing