Can GenAI actually help build better learning modules?

Can GenAI actually help build better learning modules?

Building a high-quality online learning module is a substantial undertaking that involves a range of tasks, such as writing and organising content, developing activities, sourcing multimedia, structuring pages for accessibility, and aligning it all with learning outcomes. At GAILE, we've been curious about where Gen AI might help in that process, and where it does not.

Blog by: Toni Jones, Lead, Education Gen AI, with contributions by Daniel Manning, Lead, Gen AI Experience and Kirsty Tod, Senior Learning Designer. 
 

Gen AI can accelerate the production of learning resources, but its outputs still require careful educational judgement. We have previously explored how Gen AI can support lesson planning, learning simulations and question generation, and across these examples, a consistent pattern has emerged. Without expert guidance, AI-generated content can be fluent yet generic. The educator therefore remains central to shaping outputs through pedagogical knowledge, critical review and instructional expertise.

What does the research tell us?

A related question is whether Gen AI can support the development of online content itself, specifically the material embedded in the Learning Management System, such as text, activities, assessment briefs, visuals and videos. Emerging literature on this topic indicates educators and course developers are exploring a range of applications, including improving text readability (Norberg et al., 2025), creating supporting textual and visual material (Haugsbakken, Hagelia & Nagel, 2025), and generating entire online modules (Rouabhia, 2024; Leiker et al., 2023).

The primary advantage of AI-driven course generation is the drastic reduction in time and resources required to move from a syllabus to a complete learning experience. One study generated a comprehensive, original 87-page course on multimedia databases, including 10 chapters, practical assignments, and an exam, in under a day (Rouabhia, 2024). Another study created a course 25 times faster than traditional instructional design methods (Leiker et al., 2023). The numbers are impressive, but the real story lies in how they were achieved, and that is where the available research often falls short.

An interesting account from McGregor (2026) details his iterative process with large language models (LLMs) to develop a course. Rather than accepting AI outputs at face value, he used educational theories such as Kolb’s experiential learning theory (1984) and Vygotsky’s Zone of Proximal Development as criteria for assessing, adapting and sometimes setting aside AI-generated content. One challenge he encountered was AI’s tendency to agree with him. Rather than offering alternative perspectives or constructive critique, it often reinforced his thinking, creating the conditions for an educational echo chamber. Similarly, Norberg and colleagues found that LLMs added false information or "nonsense words" when revising text, requiring constant human vigilance during quality assurance. Both examples highlight the need for users to be aware of Gen AI limitations. 

Despite advances in AI models that make them easier to use, most studies found that output quality still depends heavily on prompt specificity and an intuition for the LLM’s capabilities. Inaccurate or poorly structured prompts result in suboptimal educational material. 

An educator-led workflow in action

AT RMIT, a practical example of exploring and using Gen AI in content creation is Build a Blended Learning Module (BABL), a structured, AI-supported workflow developed by RMIT's College of Vocational Education. This repeatable process takes existing materials (PowerPoints, Word documents) and turns them into a Canvas-ready blended module. The workflow uses carefully designed prompts and templates to draft content, activities, multimedia suggestions, and accessible page structures, while educators stay firmly in charge of subject expertise, compliance, and teaching decisions; it removes repetitive assembly work.

BABL also embeds Universal Design for Learning (UDL) principles from the outset. It supports multiple ways for students to engage with content through text, video, imagery and interactive elements. It also promotes accessible design through screen-reader-friendly structures, detailed image descriptions and chunked page layouts. Because modules are developed using a consistent approach, students experience greater consistency across a program. Since November 2025, more than 100 courses developed using this workflow have achieved scores above 90% through the Canvas Accessibility measure.

To scale adoption, the College runs a ’Day of Magic’: a one-day workshop where program teams build a module together in the morning, then build their own in the afternoon. The Diploma of Business team, which attended the inaugural session in November 2025, has since rebuilt their Canvas courses using it. Modules that previously required several days of development, including creating and sourcing assets and interactive elements, can now be produced in about six hours.

Questions remain about where Gen AI genuinely adds value to content creation and where claims outpace available evidence. But BABL is a strong early example of what a well-designed, educator-led workflow can look like at scale.

What this all highlights is that there is still much to learn. I think we can say confidently that Gen AI can support the speed and quality of content creation. But the role of educators, SMEs and Learning Designers are not diminished in this process but rather are the drivers at the wheel. They need to know where they are going, who they are bringing along, and be able to confidently navigate the environment by drawing upon their knowledge of best practice.

The questions we still need to answer

What is beginning to emerge from research is that the value of AI-generated learning materials cannot be judged solely by how quickly they are produced, but by how students experience and learn from them. Early findings suggest that students do not value all AI-generated resources equally. In one small-scale study, students consistently rated AI-generated visual resources, such as images and videos, lower than learning materials. Decorative AI-generated images were often seen as less central to learning, while producing high quality educational video remained challenging (Haugsbakken, Hagelia, & Nagel, 2025). Interestingly, students placed greater value on well-designed textual resources, suggesting that clarity, relevance and pedagogical alignment matter more than the novelty of AI-generated media.

The relationship between AI-generated content and learning outcomes is similarly complex. In adaptive mathematics, students working with AI-revised problems generally achieved higher rates of mastery. However, these benefits were not consistent across all content areas, and in some cases error rates increased (Norberg et al., 2025). This highlights an important tension: while AI can make content more accessible and easier to understand, excessive simplification may reduce the cognitive effort required for learning. As a result, students may become overconfident in their understanding or engage less deeply with the material through what researchers describe as "active reading" (Norberg et al., 2025).

While the current research gives us a helpful starting point, the next step is to understand how students experience AI-supported learning materials in practice. Emerging studies are beginning to examine student perceptions, experiences and outcomes, but more work is needed to hear directly from students about what supports their learning, what feels meaningful and where AI-generated materials may fall short.

RMIT educators

If you are an RMIT educator looking to explore Gen AI within your course design and build process, the BABL workflow is available via TeachAI. You can find step-by-step instructions here, and the CoVE team are always happy to lend a hand if you need support.

In the September session of Gen AI for Educator Practice, Kirsty Tod demonstrated one of the BABL workflow processes. RMIT staff can access the session recording here: Gen AI for Educators: Emerging Practices and Possibilities

References

Haugsbakken, H., Hagelia, M., & Nagel, I. (2025). The Role of Generative AI in SPOC-Making: Student Perception of Partially AI-Generated Learning Resources. In: Hamonic, E., Sharrock, R. (eds.) Digital Education: Shaping Sustainable Lifelong Learning for All in the Era of AI. EMOOCS 2025. Lecture Notes in Computer Science, vol. 15733. Springer, Cham. https://doi.org/10.1007/978-3-032-00056-9_1

Leiker, D., Finnigan, S., Gyllen, A. R., & Cukurova, M. (2023). Prototyping the use of Large Language Models (LLMs) for adult learning content creation at scale. arXiv preprint arXiv:2306.01815. https://doi.org/10.48550/arXiv.2306.01815

McGregor, C. (2026). Generative Artificial Intelligence and Human Collaboration in Educational Design. Computer Applications in Engineering Education, 34(4), e70238. https://doi.org/10.1002/cae.70238

Norberg, K. A., Almoubayyed, H., De Ley, L., Murphy, A., Weldon, K., & Ritter, S. (2025). Rewriting content with GPT-4 to support emerging readers in adaptive mathematics software. International Journal of Artificial Intelligence in Education, 35(2), 587–626. https://doi.org/10.1007/s40593-024-00420-2

Rouabhia, D. (2024). Artificial intelligence driven course generation: A case study using chatgpt. arXiv preprint arXiv:2411.01369. https://doi.org/10.48550/arXiv.2411.01369

30 September 2026

More GAILE blogs

gen-ai-learning-material-1220x732.jpeg

Can GenAI actually help build better learning modules?
Icon / Small / CalendarCreated with Sketch. 30 Sep 2026

Building a high-quality online learning module is a substantial undertaking that involves a range of tasks, such as writing and organising content, developing activities, sourcing multimedia, structuring pages for accessibility, and aligning it all with learning outcomes. At GAILE, we've been curious about where Gen AI might help in that process, and where it does not.

four-months-in-the-lab-1220x732.jpg

Four Months in the Lab
Icon / Small / CalendarCreated with Sketch. 03 Sep 2026

One of GAILE’s primary goals: to discover the role generative AI could play a role in the curriculum; in assessment, in the student learning experience, in the review, and everywhere in between.

Diagram with transparent blocks titled: Feedback Loop, AI Interaction Trace, Reflective Checkpoint, Revision Sequence, Decision Point, Improved Draft, and Educator Judgement, representing stages in an AI-supported learning or assessment process.

Can the process be the evidence
Icon / Small / CalendarCreated with Sketch. 21 Jul 2026

The Education/Generative AI debate has reached a groundswell of debate about the struggle to secure assessment. TEQSA's latest resource, Assuring Quality Learning in a Gen AI-Integrated Future, marks a shift in a debate.

hero-pink-1220x732.jpg

Preparing Learners to Think Beyond the Output: The Role of Evaluative Judgement
Icon / Small / CalendarCreated with Sketch. 21 Jul 2026

What does it look like to teach evaluative judgement in an age of Gen AI? These examples from educators across disciplines show how students can be supported to question, test, compare, and critically evaluate AI-generated content.

Load more +
aboriginal flag float-starttorres strait flag float-start

Acknowledgement of Country

RMIT University acknowledges the people of the Woi wurrung and Boon wurrung language groups of the eastern Kulin Nation on whose unceded lands we conduct the business of the University. RMIT University respectfully acknowledges their Ancestors and Elders, past and present. RMIT also acknowledges the Traditional Custodians and their Ancestors of the lands and waters across Australia where we conduct our business - Artwork 'Sentient' by Hollie Johnson, Gunaikurnai and Monero Ngarigo.

Learn more about our commitment to Aboriginal and Torres Strait Islander peoples