Pragmatic Multimodal Teaching Using VARK and AI-Supported Pedagogy

1. Introduction

The arrival of the Fourth Industrial Revolution raised concern across many fields, and education was no exception [1, 2]. In universities and schools, people began questioning whether machines might eventually take over teaching roles [3, 4]. As AI is becoming more widely used, the message has changed. Instead of replacing educators, AI is described as a tool that could assist by taking over repetitive or routine tasks [2, 5]. Nevertheless, the application of AI in a teaching context has unearthed deep concerns about how educators work with these tools. Some educators recommend using AI to lighten administrative loads or to develop learning materials more efficiently [6]. Others are uncertain, especially when AI systems take over core tasks such as explanation, guidance, or judgement [5]. For many, the worry is not about being replaced entirely but about maintaining a clear and meaningful role in the classroom environment [5, 7]. As AI tools become more widely available, the question is not whether AI should be used in the classroom but how educators can lead in designing and curating AI-supported learning experiences.

At the same time, students have changed. Classrooms are not filled with students with the same learning preferences or abilities. Instead, students arrive with various learning needs and expectations [8, 9]. This variety in learning preferences places increasing demands on teachers. The VARK model offers a way to understand these differences [10]. Despite this, applying VARK in practice across large or diverse classes can be challenging as it requires time, flexibility and resources that are not always available to an educator [2, 8].

This is where digital tools such as AI can be helpful. With some planning, educators can present material in different formats that match various learning preferences [11, 12]. This does not mean giving over control to machines. Instead, it suggests building on the teacher’s expertise in content creation to make learning more accessible and adaptable. Combined with principles of inclusive design, this method may help remove some of the barriers students face due to language, attention, or sensory processing differences [13]. AI may not improve learning outcomes for everyone. Nonetheless, it allows for more entry points into the learning material, which, if curated correctly, may help reduce educational inequality [14, 15].

Much of today’s education debate highlights the need for more student-centred learning [8]. While this view brings important perspectives, it can sometimes underplay the value of teacher expertise in shaping learning. At the same time, approaches that rely too heavily on teacher control may become too rigid [16]. A better approach may be to shift between structured support and open-ended exploration depending on the topic and the needs of students [5]. AI can make these shifts in pedagogy easier without requiring a complete redesign of the course [10, 17]. In this pragmatic framework, the educator remains pivotal to the process. They decide what should be taught, how it should be ordered and how it connects to the broader goals of the course. These decisions should be grounded in teaching experience and understanding the subject and context. Once the design is in place, AI tools can be used to expand how that content is delivered, to support student-centred learning. For this reason, AI tools do not replace the teacher’s role but allow their expertise to be extended across different formats, helping students engage with material in ways that suit their learning preferences and pace [5, 18].

The framework suggested above is not tied to one teaching method. It creates room for movement between teacher-led guidance and student-led exploration, depending on what the situation calls for. This adaptability is supported by technology, which makes these shifts easier to manage without compromising structure. While the framework draws from ideas found in blended learning, it emphasises the teacher as the learning designer who shapes both the content and the way students interact with the content underpinned by a student-centred ethos [19]. It also supports flexible teaching strategies that respond to changing classroom needs rather than following a fixed or predetermined pattern [3].

This chapter explores how AI-supported teaching can help make the VARK model more usable in higher education. It outlines a framework where the educator remains responsible for planning and content while using AI to support more flexible and inclusive delivery. The chapter reflects on how this approach works in theory and practice, drawing from the author’s classroom. Although no general conclusions are claimed, the discussion demonstrates how teacher-guided learning, supported by AI, can be used in ways that respond more effectively to student needs and outcomes.

2. The educator’s role in the AI-enhanced classroom

The evolving education landscape in the AI era has positioned the educator in a multifaceted role that transcends traditional content delivery [20]. Rather than replacing human educators, teachers are recognised as irreplaceable due to their unique capacity to think critically and creatively, demonstrate emotional intelligence and their ability to develop social and emotional competencies in students [3]. In an AI-enhanced teaching environment, an educator becomes a guide and mentor, nurturing human connection and providing emotional support and real-world context that AI cannot replicate [4]. This shows the significance of the human touch as a pedagogical necessity provided by educators. In an AI-enabled environment, an educator’s primary function shifts towards facilitating meaningful learning experiences that leverage AI as an augmenting tool rather than a substitute for human interaction [20]. This reframing of the educator as an ‘interpersonal architect’ rather than an information provider is a significant conceptual shift. Yet surprisingly, the educator’s role as an information provider, which is a hallmark of more teacher-centred pedagogies, is regaining importance in the age of AI.

Educators must become AI literate in themselves and develop this competency in their students by enabling them to critically evaluate and ethically take advantage of AI technologies [19] beyond just technical fluency. Educators must now curate, validate and contextualise AI-produced content to ensure it aligns with learning outcomes and supports meaningful educational goals. This emerging function places educators at the heart of a complex learning environment where sound judgement and ethical guidance are essential [21]. Thus, educators play an important role in co-shaping AI’s role in educational contexts. For example, this could involve designing innovative pedagogies and assessments using AI for personalised learning, feedback and automation of routine tasks [3]. Furthermore, educators should simultaneously address concerns around academic integrity, over-reliance and developing essential skills [21] as these concerns materialise in lecture halls and classrooms. Thus, continuous professional development is necessary for educators to keep up with AI and use it to support their teaching and prepare students for an AI-driven future [19].

The preceding discussion shows that the educator’s role is not disappearing but changing meaningfully. Teachers are there not just to support students emotionally but also to make sense of the information AI tools provide. They help decide what content is valid, reliable and fits the learning goals. This calls for clear thinking, adaptability and an understanding of how AI works and how students learn best.

3. VARK learning preferences using multimodal teaching

Multimodal teaching is a pedagogical practice that focuses on mode as a key communication component in a learning environment [22]. In multimodal teaching, educators use a range of resources and tools in a pedagogical manner to enhance a student’s experience [23]. Mode, in turn, can be understood as a means of articulation that carries meaning [22]. Since all classroom communication combines multiple modes, all acts of learning are inherently multimodal [22, 24]. Learners engage with modes differently; thus, depending on a student’s competency, specific modes can enhance a student’s understanding of concepts [22]. For example, concepts in science can be taught using various modes, such as drawings, particle models and atomic models of matter to signify a concept and support a student’s comprehension from multiple angles [24]. Consequently, multimodal teaching complements the VARK framework by operationalising its core premises and by aligning with students’ preferred learning styles.

The VARK framework, championed by Neil Fleming, posits that learners have diverse learning styles, which aids in processing and retaining content [25]. Learning styles may include Visual (V), Auditory (A), Reading/Writing (R) and Kinesthetic (K) modalities, encapsulating VARK and providing a framework for analysing multimodal pedagogy [26, 27]. For example, some students may learn best through visual aids; others may prefer listening or reading, while others may prefer being actively involved in the learning process [25]. Studies have suggested that effective instructors deliver their teaching to accommodate a variety of learning styles [26, 28]. However, others have cautioned against rigidly applying learning styles, as limited evidence supports the notion that teaching to those preferences improves learning outcomes and may limit a student’s growth if not exposed to other modalities [29, 30]. Thus, we suggest following a pragmatic approach when using VARK for multimodal teaching [31] rather than strictly labelling students according to learning styles.

Multimodal teaching encourages presenting learning material in multiple ways to benefit all students regardless of their preferred learning style [32]. Employing this strategy also aligns with Universal Design principles, as numerous representations of teaching modalities may reach diverse learners through various entry points and learning channels in understanding the material [26, 33]. Thus, it is necessary to reconceptualise how technology can be used within a VARK multimodal teaching and learning environment for diversifying and consuming content [25, 31].

As previously indicated, AI opens new opportunities when thinking about VARK to personalise and adapt content across modalities for each student [34, 35]. In a traditional learning environment, a single educator may have struggled to provide individualised multimodal experiences in a large class [35]. However, AI-enhanced educational systems allow educators to scale VARK by delivering tailored content in line with student needs. A recent study by [34] on the adoption of innovative teaching approaches, specifically on the positive aspects of using AI for personalised learning to improve student engagement and performance within a self-determination theory perspective, has found and concluded the following:

• AI-enabled personalised learning promoted student engagement by positively influencing their psychological need for autonomy, competence and relatedness.

• AI contributed to measurable improvements in student outcomes as students experienced a moderate to significant improvement in their academic performance.

• AI could personalise learning through real-time data and feedback by analysing individual student performance and providing feedback and adaptive learning paths.

In effect, the study above demonstrates how AI could allow for scaling a variety of modalities not commonly found in traditional pedagogy while improving student outcomes. Students can choose how they engage with learning material and therefore not be pigeonholed into learning using a presumed learning style [29]. This blended approach, while reliant on AI-powered tools, still, however, requires the educator to orchestrate the learning environment by ensuring structured, cohesive and module outcome-relevant content is provided to students, who then can explore and construct knowledge using a variety of modalities within the VARK framework.

4. Classroom case study

The case study demonstrates the educator’s application of a pragmatic teaching approach, where AI-supported technology delivers structured, multimodal content aligned with diverse student learning needs. The participants in this case study were 127 fourth year students enrolled in the Internal Audit specialisation within a module on Advanced Organisational Governance. All students were from the same institution and engaged with the multimodal, AI supported teaching in a hybrid format as part of their regular coursework; thus, the students were not selected through a formal sampling process rather they represent the entire cohort enrolled during the semester. While these students were not research subjects in the formal sense, their interaction with the instructional design provided practical insight into its application in a real-world higher education environment. AI content creation tools were strategically employed to enhance scalability without placing an excessive workload on the educator. Notably, the educator’s role remained central in curating, validating and integrating these resources into a cohesive teaching plan. The outcome was a multimodal learning design grounded in VARK theory and pragmatism, made practical through AI-supported technologies.

The following sections describe each component of this multimodal approach. Figure 1 illustrates a framework where educators, AI tools and students interact dynamically to support multimodal teaching aligned with VARK learning preferences. At the framework’s core is the idea that AI enhances but does not replace the educator’s role and serves to personalise student learning experiences. The educator initiates the teaching process by designing and delivering content, using their pedagogical expertise to prepare material for AI-supported multimodal adaptation. AI is a central engine that transforms educator-provided inputs into multimodal formats aligned with VARK learning preferences. Students engage with the AI-generated content according to their learning preferences and provide feedback that informs further personalisation. The multimodal content produced through the collaboration between the AI and educator includes various learning formats to support learning preferences. The educator must continuously monitor student feedback to refine content delivery, which ensures human oversight throughout the learning process. This framework is suggested not to solve learning theory but to provide an educator-led approach to multimodal teaching that’s achievable and AI-augmented.

4.2 Video learning material

Having established a base for the resource material, as discussed in the planning process, the educator prepared lecture content through video presentations featuring an AI-generated avatar as the lecturer. This mode combined both visual elements and auditory narration. Video lectures can be more engaging than text alone, as they mimic a human presence and make abstract concepts more concrete through imagery. The educator was not concerned about limiting the length of videos despite this being raised as a concern in the literature [36], as students received these videos as part of their learning resources and could navigate the content according to their own pace, pause to reflect on visuals or use captions that are advantages that support self-paced learning and repetition for reinforcement.

The educator used the Animaker platform to create animated avatar videos where the avatar narrated the curated lecture script and gestured in sync with the content, as shown in Figure 2. The educator selected Animaker because it is gaining wider use in educational settings, as evidenced in recent studies suggesting that teachers and students receive the software well as it contributes valuable and versatile digital learning material to enhance learning experiences and outcomes [37, 38]. For Instance, reference [39] reports several advantages for students engaging with Animaker: simplified comprehension, improved retention, better classroom performance, enjoyable learning and increased motivation. These findings suggest that Animaker and other similar technologies can be used to transmit knowledge effectively.

From a pedagogical point of view, studies report that students perceive AI-generated content positively by enhancing the instructor’s perceived social presence, appreciating the clear and professional speech delivered by the avatar and increasing their motivation to learn, despite some instances reported of the avatars lacking naturalness in appearance and motion [40]. Conversely, other studies found no significant differences between a synthetic avatar spokesperson and an actual human spokesperson in terms of actual effectiveness and actual modality comparisons [41]. However, these findings do not align with the study by reference [36], as students interacted more with human instructor videos than with AI-generated instructors lacking human likeness. Nonetheless, instructor type did not significantly reduce academic performance. Interestingly, reference [42] reports that AI, particularly virtual reality, chatbots and sentiment analysis tools, had benefits in providing students with personalised support, increased engagement and empathy development regarding social-emotional learning in educational settings.

4.3 Audio learning material

In addition to the video, the lecture audio was extracted from the Animaker videos using standard multimedia processing software and provided as a standalone resource to students for downloading or streaming, as shown in Figure 3. This option catered directly to students who absorb information best by listening. While the videos already include audio, offering a dedicated audio format allowed students to engage in mobile learning and review content in environments where watching a screen was impossible. Pedagogically, audio in the form of asynchronous podcasts improved student regulation, engagement and learning outcomes, leading to long-term gains in student performance [43]. Students were found to prefer audio lectures over traditional face-to-face lectures due to the flexibility offered for choice for integrating learning opportunities into their daily routine [44]. Students could also play downloaded audio lectures on demand to support revision strategies, further improving learning outcomes [44].

4.4 Reading/writing learning material

The third learning modality provided to students consisted of the full-text transcripts for each lecture. The text transcripts were compiled into a learning guide, along with the corresponding PowerPoint slides, as indicated in Figure 4. The text-transcripts used were those directly created, as described in Section 4.1. The written content served three purposes; firstly, it aided students who learn best through reading, allowing them to study the lecture in text format at their own pace. Secondly, it reinforced learning by providing a reference document they could highlight, annotate and review, and thirdly, it improved accessibility for students with hearing impairments or who were non-native English speakers. Thus, students could consume the module content without watching or listening to other support modalities. Research studies indicate that educators must teach purposefully by understanding the variables that affect student learning, such as reading resources, ensuring equitable educational opportunities to support student strengths and needs [45]. While broader, reading for recreational purposes was found to have positive associations with cognitive development and academic achievement [46], which may affect reading in an educational capacity. Thus, the text-transcripts were central to reinforcing learning through reading.

4.5 Kinesthetic learning material

The fourth and last component of the educators teaching practice targeted the kinesthetic modality by incorporating practical, hands-on activities to accompany each content unit. This modality relied the least on AI support. However, the learning opportunities were supported by AI-supported content. For example, weekly practical in-person classes using traditional lecture formats were held with students. In these sessions, the students initially completed case study exam questions under exam conditions; thereafter, there was an extensive discussion between the students and the educator regarding the case study and proposed solutions. This enhanced student-centred learning as students took ownership of their learning and discussed their understanding with their classmates and professor.

As mentioned in Section 4.2, while students were given video lectures, these videos were further enhanced to support student learning, as interactive questions were embedded into the learning content; thus, as students consumed the lecture content, they were also required to complete questions before they were allowed to continue watching the content, refer to Figure 5. Examples of other activities included completing weekly online assignments covering the content reviewed for the week, submitting weekly homework based on module topics covered, and completing question banks to reinforce learning. The Kinesthetic learning experiences were designed to support active learning where concepts are applied in practice.

Literature supports the notion that kinaesthetic learners benefit from active, hands-on learning experiences and suggests that educational strategies should incorporate learning-by-doing elements to cater to student learning preferences [9, 47, 48]. For instance, a review of 225 studies found that active learning increased student performance and lowered failure rates compared to passive lecturing [49]. These findings support the value of applied learning, which was further enhanced in an AI-supported learning environment.

5. Reflections on implementation

Looking back on implementing this AI-supported multimodal approach, what stood out most was the shift in how the educators’ role evolved over the process. While the planning stage was intensive, it laid the groundwork for a smoother and more scalable experience. Once the core materials were in place, developing video lectures using AI avatar tools became more manageable, mainly because the scripts had already been structured for clarity and consistency. Though the technology was user-friendly, attention to visual timing, flow and integration with learning objectives still required thoughtful oversight.

Audio content, derived directly from the video voice-overs, was quick to generate and provided students with a mobile-friendly alternative for engaging with the material. The reading and writing resources were easily repurposed from the prepared texts with minimal formatting effort. At the same time, active and kinesthetic learning activities were integrated naturally into weekly in-person classes and interactive video content. These enhancements added depth to the learning experience without creating additional burdens.

The most valuable outcome was the ability to release high-quality content early in the semester. This allowed the educator to spend more time engaging with students, not just delivering content but guiding their learning and supporting their needs. This gave the educator confidence that the materials were accurate, consistent and adaptable across formats, making it easier to shift from a content-delivery mindset to a more student-centred, facilitative teaching approach.

6. Conclusion

This study presented a pragmatic, AI-enhanced multimodal teaching framework grounded in VARK theory and guided by educator expertise. The study demonstrated how AI tools could help scale personalised learning experiences across different modalities while reaffirming the essential role of the educator as a learning designer, facilitator and ethical guide. The case study showed that while AI can enrich the learning experience by making content more accessible, inclusive and adaptable, it cannot replace the human insight and connection educators bring to the classroom.

Theoretically, the study contributes to the evolving conversation on AI in education by integrating VARK with a pragmatic pedagogical model. On a practical level, the study equips educators with ideas and strategies to manage classrooms more effectively using AI-supported tools. From a policy perspective, the study highlights the importance of institutional support through ongoing professional development, AI literacy training and ethical guidelines that ensure responsible and equitable use of AI. Such policies should empower educators and students alike to engage critically and confidently with AI while upholding the values of academic integrity, inclusion and meaningful learning.

Several limitations of the present study should be acknowledged. First, the study did not include direct feedback from students on their perspectives and experiences. Thus, future studies should solicit student feedback to enrich the analysis. Second, the findings are based on a single case study conducted within one institution and discipline, which may restrict their applicability to other educational contexts. Hence, future studies should incorporate broader sampling across institutions and disciplines. Third, the case study reflects a single semester’s implementation without assessing long-term impacts such as student engagement, educator workload or knowledge retention. Consequently, future studies should incorporate longitudinal data to address these aspects.

Acknowledgments

The author would like to thank Zikani Mkandawire for their valuable assistance in setting up the module that served as the foundation for the current study. The author also gratefully acknowledges the University of Johannesburg for supporting the initiative and providing the necessary resources and funding.