Abstract
Artificial Intelligence is increasingly integrated into higher education for personalized learning, assessment, and institutional governance. However, challenges of trust, equity, and faculty adoption hinder its transformative potential. This paper develops a conceptual framework for AI infrastructure in education, emphasizing three layers: trust, learning, and governance. Drawing on literature and a case study of an entrepreneurship college in the Northeastern United States, the findings suggest that explainability, equitable access, and institutional oversight are critical to enhancing both student learning and faculty confidence.
Keywords: Artificial Intelligence, Higher Education, AI Infrastructure, Trust, Governance, Institutional Learning, Learning Provenance.
Introduction
The rapid adoption of generative AI has created opportunities and challenges for universities. AI can personalize learning, automate feedback, and support institutional decision-making, but questions of trust, transparency, equity, and faculty readiness remain unresolved. Treating AI as a collection of isolated tools is not enough. Institutions need an infrastructure that connects technological capability with learning outcomes and responsible governance.
The paper asks how higher education can build AI infrastructure that improves student outcomes while maintaining institutional trust and accountability. Its answer is a three-layer model built around trust, learning, and governance, supported by the concept of Learning Provenance.
Literature review
AI in education
Research on AI in education highlights the value of adaptive learning, intelligent tutoring, automated assessment, and real-time feedback. These systems can improve engagement and reduce frustration, but unequal access and insufficient guidance can deepen existing educational disparities.
Trust and governance
Trust depends on explainability, transparency, fairness, privacy protection, and the ability to audit how AI is used. Governance therefore cannot be added after deployment; it must be part of the infrastructure through policy controls, oversight dashboards, bias monitoring, and clear institutional accountability.
Institutional learning
Universities learn when they can observe how students, faculty, and systems interact over time. AI infrastructure should help institutions identify effective learning patterns, improve teaching practice, and refine policy without reducing education to surveillance or one-off output measures.
Methodology
The study combines three methods: a review of literature on AI in education, trust, and governance; a case study deployment of Answerr AI at an entrepreneurship college in the Northeastern United States; and development of a three-layer conceptual framework based on the observed implementation.
The AI infrastructure framework
Trust layer
Explainable AI, audit logs, and bias detection make AI use transparent, fair, and reviewable. This layer establishes confidence among students, faculty, and institutional leaders.
Learning layer
Adaptive content, multi-model access, and usage records support better learning experiences. The objective is improved confidence, reduced frustration, and visibility into how learners develop ideas—not simply whether they produced an answer.
Governance layer
Data privacy compliance, oversight dashboards, and institutional ethics structures enable responsible adoption and continuous improvement. Administrators can govern access and policy while faculty retain meaningful control over learning design.
Case study: an entrepreneurship college
The participating college faced student disengagement and frustration, while faculty were concerned about inequity, plagiarism, and inconsistent access to AI. The deployment introduced multi-model access, automatic activity records, and equitable usage within a governed environment.
Students reported that guided AI access reduced the anxiety of getting stuck and helped them compare different ways of approaching a problem. Faculty gained dashboards that showed participation and learning patterns, creating a basis for support and course improvement rather than relying on hidden detection tools.
The case suggests that successful adoption depends on more than model availability. Equitable access, visible process, clear expectations, and faculty oversight work together to build confidence and improve learning quality.
Learning Provenance
Learning Provenance is a record of how learning develops across resources, experiences, and outcomes. Resources include textbooks, faculty material, AI-generated content, and peer contributions. Experiences include study sessions, projects, drafts, AI-assisted problem-solving, and revision cycles. Outcomes include assignments, assessments, demonstrated skills, and reflection.
By connecting these elements, institutions can understand how knowledge was formed and how capability changed over time. The focus shifts from policing a final output to supporting the learning process behind it.
Discussion
The framework positions AI infrastructure as a shared institutional capability. Trust makes adoption credible, the learning layer makes it educationally useful, and governance makes it sustainable. Removing any one of these layers weakens the whole system.
Conclusion
Higher education should move beyond isolated AI tools toward infrastructure that is explainable, equitable, measurable, and institutionally governed. The three-layer framework and Learning Provenance provide a practical foundation for improving student learning and faculty confidence while protecting accountability and trust.
Authors
Mohd Qaiser Malik, MBA’25 Babson (author), and Trond Undheim, PhD (co-author). The complete paper includes the research references and supporting case-study figures.