Towards a Flipped Learning Ecosystem: A Generative Artificial Intelligence-Enabled Framework

Flipped classroom as a mode of blended learning

Alqarni (2018) classified blended learning models into three categories: low-impact (integrating online activities into traditional teaching), medium-impact (restructuring a course by substituting some face-to-face approaches with online features), and high-impact (developing blended learning materials aligned with the curriculum). The high-impact category facilitates the seamless integration of online and face-to-face learning, allowing instructors to maximize the benefits of blended teaching and effectively cater to students’ needs. On the other hand, Alqarni (2018) classified the FC method into two categories: traditional and holistic flipped classrooms. In the conventional type, students learn by watching lecture videos, having interactive sessions, and exploring challenging topics comprehensively (Asiksoy & Ozdamli 2016). The instructor then develops activities based on the material and provides face-to-face support.

To establish the flipped classroom as a significant type of blended learning, it is crucial to study the design of the learning process and environment. The design plays a vital role in evaluating instructional effectiveness, including the FC approach, while emphasizing the theoretical and practical underpinnings of student-centered and active learning (Ng & Lam 2022). The holistic flipped classroom approach considers every learning activity a classroom affair, supported and monitored by the instructor (Asiksoy & Ozdamli 2016).

Therefore, optimizing the impact of the blended learning approach or its subsets, such as FL, involves revisiting the entire learning process—from curriculum structure to learning design—rather than merely outsourcing part of the teaching through videos. In Figure 1, the traditional flipped classroom practice norm remains in the medium-impact blended learning category, involving decisions about which parts of the course to flip and which not to flip. In contrast, the holistic flipped classroom emphasizes a high-impact category of blended learning, and advocating for this approach remains one of the aims of this paper.

Figure 1

The benefits of the flipped approach

The FCM enhances learning by focusing on student-driven activities and personalized support (Bergmann & Sams 2012). As enrollment in higher education rises, reforming traditional teaching methods becomes essential. This model fosters higher-order thinking, communication, and self-directed learning (Bishop & Verleger 2013; Bergmann & Sams 2012). Effective implementation requires students to collaborate and actively build knowledge, leading to deeper insights. The FC is now a vital teaching approach that has a positive impact on educational experiences (Bishop & Verleger 2013). In a qualitative discussion with expert professors using the FC method in their classes over multiple years, an account of the most important benefits have been compiled (See Table 1; ElGamal, 2024).

The opportunities of GenAI for flipped learning

AI has the potential to transform education, making the role of educators and institutions essential in guiding its responsible development. Systematic reviews, such as the one by Zawacki-Richter et al. (2019), identify key AI applications in education, which include personalized learning, automated assessment, predictive analytics, and intelligent tutoring. These applications align with the Framework for Learning Models (FLM) by promoting personalized and engaging learning experiences.

Xu and Ouyang (2022) further highlight AI’s diverse roles: it can serve as a subject of study, a direct learning tool, and an assistant for both students and educators. This multifaceted approach enhances interaction between students and teachers, facilitates peer collaboration, and empowers self-directed learning. In doing so, it effectively addresses key challenges faced in blended learning environments.

GenAI is a subset of AI that creates content in various modalities and media and has the potential to revolutionize education. Models like Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and Transformers, including Generative Pre-trained Transformers (GPT) and Gemini, power this advancement (Alshahrani 2023). The integration of transformer-based models, including Generative Pre-trained Transformers (GPT) and submissions from Gemini, has contributed significantly to the current enthusiasm surrounding GenAI. These models use machine learning to respond to queries and generate images, text, and videos. Moreover, the combination of machine learning, natural language processing, and chatbot technologies has the potential to significantly enhance various aspects of education, including content delivery, assessment, and feedback (Alshahrani 2023). These models enable personalized learning, automated content generation, and data-driven insights. By tailoring assignments to individual needs, analyzing student performance, and automating tasks, GenAI enhances both teaching and learning experiences (Sengar et al. 2024).

The increasing interest in research on GenAI in education has led to several systematic literature reviews. Notable analyses by Sengar et al. (2024) and Ipek et al. (2023) have highlighted the perceptions of AI-powered chatbots among educators and learners in educational settings, particularly in blended learning environments. A consistent trend has emerged: learners are enthusiastic about using these tools, while educators tend to be more cautious and skeptical.

Recent investigations (Lin & Mubarok 2021; Al-Kaisi et al. 2021) into the use of GenAI within FL frameworks have provided policymakers and educational professionals with valuable insights. These insights focus on benefits such as resource optimization and enhanced educational accessibility in blended learning settings. Current research is shifting from merely assessing the effectiveness of FL compared to traditional teaching methods (ElGamal 2022) to exploring innovative strategies and technologies that enhance foreign language learning. This shift opens up new and impactful research opportunities.

GenAI can potentially enhance FL by providing personalized support, automating tasks, and fostering student-educator interactions. It can integrate intelligent tutoring, data analysis, content recommendations, and virtual assistance, thereby transforming student engagement and creating inclusivity (Chiu 2024; Alshahrani 2023). Tools like ChatGPT improve personalized learning and motivation in larger classes, offering tailored solutions and instant responses that address low student interest and feedback delays (Li 2023; Lee et al. 2022).

A study by López-Villanueva et al. (2024) demonstrated that incorporating GenAI into flipped learning enhances student engagement, requiring instructors to adapt and acquire new skills. Ilieva et al. (2023) proposed a five-stage framework for integrating GenAI-driven chatbots in blended learning, tackling learner engagement and workload issues. Research by López-Villanueva and colleagues (2024) showed GenAI’s effectiveness in flipped learning by providing tailored resources and ongoing student support (Sengar et al. 2024). Alshahrani (2023) identified strategies for implementing GenAI that enhance accessibility and educational equity, emphasizing the importance of careful integration and teacher training. Sánchez-Ruiz et al. (2023) highlighted ChatGPT’s role as a supplementary tool in engineering education.

Finally, Park and Doo (2024) called for research to consider emotional responses and motivation alongside learning outcomes. In summary, GenAI has the potential to enhance social interaction, support learning, and foster emotional aspects in blended and flipped learning environments.

Research questions

1. How can the strategic integration of GenAI tools optimize the implementation and effectiveness of FL at the meso (institutional support) and micro (individual learning) levels in higher education?

2. How can a connectivist-informed framework integrating GenAI optimize in-class activities within flipped learning pedagogy to foster deeper learning and collaborative knowledge construction in higher education?

Conceptual research design

Following Jaakkola’s (2020) typology, this study adopts the model approach, which seeks to build a conceptual framework that clarifies relationships between constructs and develops propositions for future exploration. The conceptual framework that will be introduced in the coming section is thus not merely descriptive but explanatory, positioning FL as a multi-level ecosystem reshaped by GenAI.

Literature identification and synthesis

The framework is grounded in an umbrella review of FL in higher education (ElGamal & Zawacki-Richter, 2025), which categorizes barriers into student-related, faculty-related, and operational domains guided by Betihavas et al. (2016). The systematic reviews included in this umbrella review were published between 2018 and 2022 and sourced from three major international databases—Web of Science, Education Source, and Scopus. Each review was a peer-reviewed scholarly journal article, published in English, with FL as its primary focus, and within the defined time span, which constituted the unit of analysis. This foundation was further enriched with additional literature selected specifically for its conceptual and theoretical relevance, in line with best practices for conceptual research (Gilson & Goldberg 2015).

Model-building process

The analysis proceeded in three iterative stages (Jaakkola, 2020):

1. Deconstruction – identifying key constructs and themes from umbrella and complementary studies.

2. Integration – organizing these constructs through the dual lens of connectivist theory and GenAI affordances at the micro, meso, and macro levels.

3. Framework Construction – synthesizing the findings into a systemic conceptual framework or model that illustrates the dynamic interplay of learners, faculty, institutions, and theoretical paradigms in a GenAI-augmented flipped learning environment.

The FL challenges

An umbrella review of the literature underscores these complexities, categorizing them into student-related, faculty-related, and operational challenges that demand careful consideration and strategic solutions (ElGamal & Zawacki-Richter 2025, Forthcoming).

The proposed conceptual framework

Given the intricate challenges in implementing FL, a comprehensive, systemic approach is essential (Wang et al. 2015). This conceptual framework introduces the GenAI-Enabled Flipped Framework (GEFF), which extends existing theoretical models by integrating the transformative capabilities of GenAI. GenAI-Enabled Flipped Framework positions FL not as an isolated pedagogical method but as an integral component of a dynamic learning ecosystem—a complex network of learners, educators, content, and technology that co-evolves to maximize educational impact (Dillenbourg & Jermann 2010; Giannakos et al. 2016; Siemens 2005).

This framework defines a learning ecosystem as a diverse, collaborative, and dynamic network of stakeholders that provides greater access to a range of learning opportunities and supports learners in achieving positive academic and wellbeing outcomes (Dillenbourg & Jermann 2010). This systemic view is the foundation for GEFF, which is grounded in three key theoretical perspectives:

• Learning Ecosystem Concept. The framework adopts a holistic view of education, recognizing that FL is not an isolated technique but part of a larger system of interactions and dependencies (Dillenbourg & Jermann 2010; Giannakos et al. 2016; Siemens 2005). GenAI-Enabled Flipped Framework (GEFF) builds upon Wang et al.’s (2015) Complex Adaptive Blended Learning Systems (CABLS) framework by explicitly including GenAI as a key component. Six interconnected subsystems are identified by CABLS: learner, teacher, institution, content, learning support, and technology. The dynamic and non-linear relationships among these elements, which are now influenced by GenAI, are recognized by GEFF.

• Holistic Learning Drivers. GenAI-Enabled Flipped Framework (GEFF) aligns with the ‘Classroom 4.0’ vision, which is driven by engaging pedagogies like FCs and e-learning (Koul & Nayar 2020). It operationalizes this vision by embedding GenAI-driven personalization and adaptability into flipped practices. Existing frameworks lack explicit consideration of GenAI’s potential in optimizing FL (Park & Doo 2024), a gap that GEFF fills.

• The Complex Adaptive Blended Learning Systems (CABLS) framework. Wang and colleagues’ (2015) Framework CABLS provided a pivotal contribution to our systemic understanding of blended learning. Its examination of six interconnected dimensions (learner, teacher, institution, content, learning support, and technology) effectively illuminated the dynamic and non-linear interactions within these environments. Nevertheless, despite its robust analysis of blended learning’s complexities, CABLS did not explicitly account for the transformative potential of emerging technologies, most notably GenAI, in optimizing flipped learning outcomes. Recognizing this gap, the framework presented in this paper builds upon the foundational work of CABLS, extending it a decade later to integrate the crucial role of GenAI.

Integration of learning theories

To ensure a robust theoretical grounding, GEFF draws on both established and contemporary learning theories, proposing a nuanced integration of two key theories: constructivism and connectivism.

(GenAI-Enabled Flipped Framework) GEFF is grounded in robust theoretical underpinnings to address the long-standing critique that research in ODDE has often been descriptive and lacking theoretical foundations (Perraton 2000; Zawacki-Richter & Bozkurt 2023). By integrating both established and contemporary learning theories, GEFF provides a holistic and future-oriented foundation for flipped learning.

At its core, GEFF draws on constructivism, emphasizing active, learner-centered engagement where knowledge is co-constructed through scaffolding, guidance, and social interaction (Vygotsky 1978; Bishop & Verleger 2013). Within flipped learning, the Zone of Proximal Development (ZPD) becomes central, highlighting the role of adaptive scaffolding, peer collaboration, and the presence of more knowledgeable others in supporting learners’ progression.

Figure 2

Complementing this, GEFF incorporates connectivism, a network-based theory that recognizes the transformative role of digital technologies in fostering diversity, autonomy, and interactivity (Siemens 2005; Downes 2010; Bates 2015). Connectivism views learning as a process of creating and navigating networks, with knowledge emerging from interactions, collaboration, and the exchange of perspectives within digital ecosystems (Kop & Hill 2008; Chien et al. 2019). It underscores the potential of technology and GenAI tools to expand opportunities for inclusive, adaptive, and interactive learning environments.

Rather than a linear transition between these theories, GEFF envisions a dynamic integration of constructivism and connectivism. While connectivist principles often inform the pre-class phase of flipped learning through self-directed exploration and digital interaction, constructivist approaches underpin the in-class phase, where collaboration and knowledge reconstruction occur (Chien et al. 2019). When mediated by GenAI, these phases converge, enabling a fluid interplay between individual exploration and collective knowledge-building.

This integration addresses key limitations of traditional FL, such as (and not limited to) students’ tendency toward isolated engagement (Lo 2022). By embedding scaffolding strategies and leveraging GenAI to foster collaboration, GEFF proposes a model that values diversity, autonomy, interactivity, and openness (Downes 2010), thereby offering a comprehensive theoretical grounding for flipped learning in the digital age.

The GEFF: Multi-level operation

GenAI-Enabled Flipped Framework (GEFF) is a modified and extended version of CABLS, structured around a 3M model (macro, meso, micro levels). The effectiveness of GEFF stems from its strategic integration of GenAI across three interconnected, dynamic levels (see Figure 3).

Figure 3

Specifically, the framework illustrates:

• GenAI as a Mediator: GenAI mediates the relationship between the teacher and content by automating content generation and personalizing it for the learner.

• Feedback Loops: The framework emphasizes continuous feedback loops, where GenAI provides dynamic, personalized feedback to the learner, which in turn informs the teacher and the institution.

• Holistic System: The framework posits that changes to one component—particularly the introduction of GenAI—will have a cascading effect on the entire ecosystem, requiring a systemic approach to implementation.

This multi-level approach (see Figure 3) not only addresses the inherent challenges of the flipped learning model but also predicts the potential impacts at each level. For researchers, GEFF offers a comprehensive framework for examining the multifaceted dimensions of flipped learning, uncovering critical connections and pathways for investigation.

The GEFF: Interrelation and dynamics

The GEFF is not a static model but a complex, adaptive, and dynamic ecosystem where changes in one component ripple across the entire system. It operates across three interconnected levels:

• Macro Level (Theoretical Advancement): The GEFF redefines FL within the broader context of learning ecosystems and GenAI-enabled knowledge networks.

• Meso Level (Institutional Integration): The framework guides policy, faculty development, and infrastructure planning for the sustainable adoption of FL.

• Micro Level (Classroom Practice): The GEFF enhances pre-class, in-class, and post-class phases through GenAI-driven personalization, collaboration, and adaptive feedback.

For researchers, GEFF provides a comprehensive framework for investigating the evolving complexities of FL. For practitioners, it offers a practical toolkit to address systemic challenges by strategically embedding GenAI to enhance engagement, optimize learning, and expand inclusivity.

GEFF: Gen-AI enabled macro-level

At the macro level, the framework articulates the overarching theoretical principles that anchor the approach, drawing primarily on social constructivist and connectivist perspectives. These paradigms shape the institutional vision and inform the design of strategies and pedagogical models at the meso and micro levels, fostering a coherent, learner-centered, and digitally integrated educational ecosystem. However, the integration of GenAI at this level warrants both optimism and caution. GenAI presents a clear opportunity for large-scale curriculum redesign through the rapid creation and adaptation of learning materials tailored to diverse learner profiles, including those with accessibility and neurodiversity considerations (Qian 2025), but its transformative potential depends on the institution’s capacity to address ethical, epistemological, and quality-assurance concerns.

Similarly, the promise of leveraging AI-driven analytics to identify educational trends, forecast skill demands, and inform policy (Qian 2025) is tempered by the risks of over-reliance on algorithmic outputs, potential bias in training data, and the erosion of academic autonomy (Cefa et al. 2025). Furthermore, while strategic GenAI adoption can strengthen the alignment between institutional goals and pedagogical practice, fostering a sustained cultural transformation (Cefa et al. 2025), such an outcome (as claimed in GEFF) would require deliberate governance, robust faculty development, and a clear framework for responsible use. Without these safeguards, the macro-level adoption of GenAI risks amplifying inequities rather than promoting inclusive, future-ready higher education.

GEFF: Gen-AI enabled meso-level

At the meso level, institutions aiming to scale FL must go beyond isolated classroom innovations and focus on systemic integration. Here, implementation for mitigation becomes the strategic goal—specifically, reducing workload, time pressures, resource gaps, and instructional inefficiencies that hinder widespread FL adoption (see Table 2). This is made possible through three interdependent pillars within the GEFF framework: centralized support systems, inventory development, and training—each enhanced by GenAI tools.

Also at the meso level, the framework focuses on the institutional and organizational capacities that bridge the macro-level pedagogical vision with micro-level learner engagement. This intermediary layer is critical for translating theoretical constructs into sustainable, scalable practices, particularly in the context of FL, where institutional support, strategic planning, and educator preparedness have been identified as recurrent implementation challenges (Akçayır & Akçayır 2018; Lo & Hew 2022; Article III). By systematically addressing these challenges through coordinated measures at the meso level, such as strengthening technology infrastructure, enhancing academic and technical support, maintaining dynamic institutional knowledge bases, and delivering targeted professional development, institutions can mitigate factors that typically hinder FL adoption. Such hindering factors at the meso level often include insufficient access to up-to-date learning technologies, lack of centralized resource management, inconsistent technical support, fragmented professional development provision, and absence of coordinated instructional planning. In GEFF, GenAI tools play an enabling role in both institutional support and institutional readiness, ensuring that meso-level functions are adaptive, data-informed, and aligned with the pedagogical goals of flipped learning.

The institutional support comprises the technological infrastructure, academic assistance systems, and curated knowledge resources that underpin effective FL implementation. In a GenAI-enabled framework, these components are integrated into a cohesive support ecosystem.

• Technology Infrastructure. Robust digital and physical learning environments (including high-speed connectivity), secure LMS access, and device compatibility form the base. GenAI enhances these through intelligent tagging, automated indexing, and predictive analytics, enabling efficient resource allocation and personalized access.

• Learning Support. GenAI-powered tutoring agents offer just-in-time, subject-specific guidance, formative feedback, and adaptive remediation. GenAI-driven help desks provide instant technical troubleshooting and platform navigation assistance, minimizing downtime. Wikis also act as collaborative, living repositories of institutional knowledge. GenAI keeps them current via automated summarization, outdated content detection, and integration of best FL practices, ensuring both faculty and students have an evolving and accessible knowledge base.

• Open Educational Resources (OERs). GenAI curates and generates OERs, adapts them to local curricula, and translates them for multilingual access. Linked with institutional wikis, these resources enhance accessibility and pedagogical richness.

On the other hand, institutional readiness focuses on the strategic and human-capital aspects necessary for implementing and sustaining FL at scale. Through intelligent planning, data-informed decision-making, and adaptive professional development, GenAI optimizes readiness.

• Teacher Training – Generative AI-personalized pathways, adaptive microlearning, and real-time mentoring simulations enhance professional development. Generative AI analyzes teaching practices to deliver targeted feedback on flipped techniques, digital pedagogy, and assessment design.

• Adaptive Planning – Generative AI generates predictive models for resource allocation, scheduling, and workload distribution, and simulates FL rollout scenarios to optimize strategies.Implementation for Mitigation Example: Poor planning overburdens staff and wastes technology; AI-driven modelling enables efficient implementation.

GEFF: Gen-AI enabled micro-level

Finally, at the micro-level, the framework addresses the immediate teaching-learning interface, where pedagogical vision is enacted in real time. Here, the teacher shifts from being a primary knowledge transmitter to a facilitator—guiding inquiry, moderating collaboration, and nurturing learner autonomy—while students actively engage as co-constructors of knowledge. In the FL context, this facilitation is strengthened by GenAI’s ability to personalize experiences, scaffold interaction, and streamline assessment, making learning more adaptive, interactive, and student-driven. Generative AI tools support this by reducing cognitive load through summarization, content retrieval, and real-time feedback, enabling more time for higher-order tasks like problem-solving, simulations, and peer instruction (Zhen & Yahya 2024).

For example, in a STEM-based module, students might engage with a GenAI-powered simulation during the in-class phase to model real-world scenarios. The GenAI tool dynamically generates visual or data-driven outcomes based on student inputs, allowing learners to experiment with variables, test hypotheses, and receive immediate, adaptive feedback. This form of micro-level facilitation reduces the cognitive load associated with abstract concepts and grounds learning in an interactive, experiential environment. This level constitutes the following elements:

• Personal Learning Environment (PLE). Generative AI personalizes learning through adaptive pathways, content recommendations, and intelligent resource discovery. Artificial intelligence dashboards integrate OERs, institutional materials, and peer contributions, enabling learners to curate their own study spaces and build self-regulation.

• Content. Generative AI generates adaptive case studies, problem scenarios, and multimedia aligned with pedagogical goals, adjusting complexity to learner progress.

• Interaction. Using connectivist principles, GenAI moderates discussions, matches peers, and summarizes dialogues, highlighting themes for deeper inquiry.

• Assessment. Generative AI delivers instant, personalized feedback, adapts future assessments, and tracks growth across learning outcomes.

At the heart of this holistic learning ecosystem is the GenAI-enhanced micro-level instructional design, aligned with connectivism’s focus on networked, adaptive learning. Using Kultawanich et al.’s (2015) four-stage model—aggregating, remixing, repurposing, and feed-forwarding—students engage in interactive cycles of inquiry, creation, and peer feedback. This micro-level facilitation transforms the in-class experience into a space for dynamic knowledge construction. GenAI tools do not replace pedagogical intent but enhance it, enabling precision, personalization, and engagement at the activity level. The result is a more coherent, emotionally connected, and cognitively supported learning experience for both traditional and digital classrooms.