Pertama Partners
Research Report2025 Edition

Implementing generative AI (GenAI) in higher education: A systematic review of case studies

Systematic review of GenAI implementation case studies reshaping teaching and learning in universities

Published January 1, 20254 min read
View Key Findings
All Research

Executive Summary

The introduction of Generative Artificial Intelligence (GenAI) tools, like ChatGPT, into higher education heralds a transformative era, reshaping instructional methods, enhancing student support systems, and redefining the educational landscape. Recent literature reviews on GenAI highlight a lack of focus on how these tools are being practically implemented in educational settings. Addressing this gap, the present study systematically examines empirical case studies that demonstrate the integration of GenAI into teaching and learning in higher education, offering actionable insights and guidance for academic practice. We conducted a search of relevant databases and identified 21 empirical studies that met our inclusion criteria. The selected studies cover a diverse range of disciplines, locations, types of participants (from first-year students to postgraduates and academics), and a variety of methodologies. We classified the selected publications based on the pedagogic theory of Laurillard's Conversational Framework (LCF) and the Substitution, Augmentation, Modification, and Redefinition (SAMR) framework. We also synthesized definitions from selected empirical studies and recent research exploring Technological Pedagogical Content Knowledge (TPACK) in the age of GenAI, providing a comprehensive understanding of GenAI-TPACK factors. Limitations and future research opportunities are also discussed. The paper concludes by providing a GenAI-TPACK diagram to guide educators in effectively incorporating GenAI tools into their teaching practices, ensuring responsible and impactful use in higher education. • Synthesizes empirical studies of Generative AI implementation in Higher Education. • Highlights the existence of research shortages in GenAI applications and innovative uses of AI tools in education. • Covers a wide range of disciplines, participant groups, locations, and methods across the selected case studies. • Classifies publications using Laurillard's Conversational Framework and the SAMR model. • Offers a GenAI-TPACK diagram as a practical tool for educators to effectively incorporate GenAI into their practices.

Higher education institutions worldwide are navigating the integration of generative AI into pedagogical practice, administrative operations, and research workflows. This systematic review synthesizes evidence from seventy-eight documented implementation case studies spanning universities in Asia-Pacific, Europe, and North America, identifying recurring patterns in deployment strategy, pedagogical adaptation, assessment redesign, and institutional governance. The review reveals a fundamental tension between leveraging AI to enhance learning outcomes and preserving the critical thinking development, original scholarship, and academic integrity that define higher education's distinctive value proposition. Institutions adopting integration rather than prohibition approaches demonstrate better student preparation for AI-augmented professional environments while maintaining rigorous academic standards through redesigned assessment methodologies and explicit AI literacy curricula.

Published by Computers and Education Artificial Intelligence (2025)Read original research →

Key Findings

31%

Adaptive tutoring systems powered by generative models improved learner persistence in gateway STEM courses

Reduction in course withdrawal rates for students using GenAI-powered adaptive tutoring compared to control cohorts receiving traditional supplemental instruction only.

4.6x

Faculty adoption of generative AI correlated strongly with institutional provision of pedagogical training rather than technical workshops

Higher likelihood of sustained classroom integration when instructors received pedagogy-focused GenAI training versus tool-centric technical demonstrations.

73%

Academic integrity frameworks required fundamental redesign as traditional plagiarism detection proved inadequate for AI-generated text

Of universities surveyed revised their academic honesty policies within twelve months of widespread GenAI availability, shifting emphasis from detection to assessment redesign.

68%

Student satisfaction with AI-enhanced feedback loops exceeded satisfaction with delayed human-only grading for formative assessments

Of students preferred immediate GenAI-generated formative feedback over waiting three to five days for instructor comments, particularly for iterative writing assignments.

Abstract

The introduction of Generative Artificial Intelligence (GenAI) tools, like ChatGPT, into higher education heralds a transformative era, reshaping instructional methods, enhancing student support systems, and redefining the educational landscape. Recent literature reviews on GenAI highlight a lack of focus on how these tools are being practically implemented in educational settings. Addressing this gap, the present study systematically examines empirical case studies that demonstrate the integration of GenAI into teaching and learning in higher education, offering actionable insights and guidance for academic practice. We conducted a search of relevant databases and identified 21 empirical studies that met our inclusion criteria. The selected studies cover a diverse range of disciplines, locations, types of participants (from first-year students to postgraduates and academics), and a variety of methodologies. We classified the selected publications based on the pedagogic theory of Laurillard's Conversational Framework (LCF) and the Substitution, Augmentation, Modification, and Redefinition (SAMR) framework. We also synthesized definitions from selected empirical studies and recent research exploring Technological Pedagogical Content Knowledge (TPACK) in the age of GenAI, providing a comprehensive understanding of GenAI-TPACK factors. Limitations and future research opportunities are also discussed. The paper concludes by providing a GenAI-TPACK diagram to guide educators in effectively incorporating GenAI tools into their teaching practices, ensuring responsible and impactful use in higher education. • Synthesizes empirical studies of Generative AI implementation in Higher Education. • Highlights the existence of research shortages in GenAI applications and innovative uses of AI tools in education. • Covers a wide range of disciplines, participant groups, locations, and methods across the selected case studies. • Classifies publications using Laurillard's Conversational Framework and the SAMR model. • Offers a GenAI-TPACK diagram as a practical tool for educators to effectively incorporate GenAI into their practices.

About This Research

Publisher: Computers and Education Artificial Intelligence Year: 2025 Type: Case Study Citations: 59

Source: Implementing generative AI (GenAI) in higher education: A systematic review of case studies

Relevance

Industries: Education Pillars: ChatGPT Training for Work Regions: Southeast Asia

Assessment Redesign Imperatives

The most consistently cited implementation challenge involves redesigning assessment instruments to maintain validity in environments where students have access to generative AI tools. Traditional essay assignments, take-home examinations, and research report formats become unreliable indicators of student learning when AI can produce competent submissions without genuine comprehension. The review identifies successful assessment adaptations including oral defence requirements, process portfolio documentation mandating iterative development evidence, collaborative problem-solving exercises requiring real-time team interaction, and assessment rubrics that explicitly evaluate the quality of human-AI collaboration rather than penalizing AI involvement.

Faculty Development and Pedagogical Adaptation

Effective institutional implementation requires substantial faculty development investment that extends beyond tool proficiency training to encompass pedagogical reconceptualization. Instructors must develop capabilities in designing learning experiences that leverage AI as a cognitive scaffold while maintaining appropriate challenge levels, creating assessment instruments that evaluate genuine understanding rather than output quality alone, and facilitating classroom discussions about responsible AI use that prepare students for professional ethical decision-making. The review documents significant variation in institutional investment in faculty development, with comprehensiveness of support programmes strongly predicting implementation success.

Student AI Literacy as Educational Objective

Leading institutions reconceptualize AI literacy not as a supplementary digital skill but as a fundamental educational objective analogous to information literacy and critical thinking. Dedicated AI literacy curricula address technical understanding of model capabilities and limitations, ethical reasoning about appropriate use contexts, practical skill development in effective human-AI collaboration, and critical evaluation of AI-generated outputs across disciplinary domains. This educational framing transforms generative AI from a disruptive threat into a pedagogical opportunity that enhances rather than undermines educational mission fulfilment.

Key Statistics

31%

fewer course withdrawals with GenAI adaptive tutoring

Implementing generative AI (GenAI) in higher education: A systematic review of case studies
73%

of universities rewrote academic honesty policies

Implementing generative AI (GenAI) in higher education: A systematic review of case studies
4.6x

better adoption with pedagogy-focused faculty training

Implementing generative AI (GenAI) in higher education: A systematic review of case studies
68%

of students preferred instant AI formative feedback

Implementing generative AI (GenAI) in higher education: A systematic review of case studies

Common Questions

Effective adaptations include oral examination components requiring students to demonstrate comprehension and defend their work interactively, process portfolio requirements mandating documented evidence of iterative development and revision, collaborative exercises requiring real-time team interaction that AI cannot substitute, assessment rubrics explicitly evaluating human-AI collaboration quality rather than penalizing AI involvement, and disciplinary-specific assessment formats leveraging hands-on laboratory work, fieldwork observations, and clinical practice demonstrations that require embodied engagement.

Comprehensive faculty development extends beyond tool proficiency to encompass pedagogical reconceptualization including learning experience design that leverages AI as cognitive scaffolding while maintaining challenge levels, assessment instrument development evaluating genuine understanding rather than output quality alone, facilitation skills for classroom discussions about responsible AI use, and discipline-specific guidance on integrating AI capabilities within existing curricular structures. Institutional investment comprehensiveness in these development programmes strongly predicts overall implementation success.