Starting this project was challenging. I had an idea in mind, to create something useful and meaning, but once I realised the full scope of the project, all the documentation required, the steps I would have to follow and the unknowns I would have learn about, I was overwhelmed. Looking back, I realised I had no real grasp of what the full projected entailed, and I was often lost in theory, looking back at my notes, re-reading chapters of books or articles, to find a way through. It was then that it occurred to me: let me start with what I know, and who I know. I started with the learners, reviewing the learning personas I had created and the data I had collected.
An initial consultation with the International English Library's Committee identified the business need for a basic AI literacy course. As the library operates on a 100% voluntary basis, effective use of AI could reduce time spent on administrative tasks and streamline processes, while also sparking general interest from volunteers to further engage with new opportunities. This resulted in the creation of a survey to gauge the volunteers' AI learning needs - the gap between where they are now and where they want to be (Knowles et al., 2025) - as well as their attitudes and experience regarding eLearning.
To structure the project, I chose the ADDIE model for its systematic and linear approach, appropriate for a project with fixed goals and for a novice designer seeking a clear framework. I recognized its rigidity could be a limitation in a volunteer-driven environment where learner needs might change; however, a defined structure was essential for managing the project's scope. So, the first challenge was the analysis. After collating the information into two personas, it became evident that the course needed to address both their needs. While it was easier to accommodate common criteria (e.g., social learning, downloadable resources), only a differentiated pathway design could address the AI knowledge learning gap. Therefore, the course was designed using a combination of universal modules (module 1 to diagnose the AI knowledge level and module 3 to conclude the course and identify further learning opportunities) and two versions of module 2: one for those who have limited AI knowledge and another for those who already have some. Google Gemini was used to summarize the main similarities and differences between the two personas. While this accelerated the data synthesis, I cross-referenced the results to interpret the nuances of the qualitative survey data and prevent the AI from oversimplifying the learners' motivations, ensuring the final personas remained authentic.
I felt that this was an effective way to gather data and support form stakeholders. In hindsight, I could have spent some time interviewing some of the volunteers who expressed the need for an AI Literacy course, to gather more detailed information. I am planning to focus on their feedback on the course to implement changes, to remedy this oversight.
Reflecting critically on the persona development process, while the survey data clearly identified two distinct user groups, I recognize that this binary approach may have oversimplified the complexity of adult learning preferences. I could have incorporated additional variables such as cultural learning preferences, particularly given the international volunteer base at the library. This could be a consideration for future projects.
By using a self-diagnostic method in module 1, flexibility was provided to learners: based on a self-assessment based on the learning outcomes, learners can choose their module 2 pathway. Additionally, they have access to the other pathways, offering an opportunity to consolidate learning, scaffold, or challenge themselves further. This survey also indicated that a blended learning experience would be most beneficial, including opportunities for face-to-face and digital peer support. It had a clear impact on the choice of modalities and assessment, as volunteers' needs demanded flexibility (optional assessments) and eLearning that was interactive, scaled down, and included basic language and support in German for some. However, I recognize that without implementation and formal evaluation, the true effectiveness of these design decisions remains theoretical. Future projects could benefit from user testing and changes based on actual learner feedback.
To identify the content of the course, extensive research was conducted on what basic AI literacy skills include, defined by Long & Magerko (2020, p.2) as 'a set of competencies that enables individuals to critically evaluate AI technology; communicate and collaborate effectively with AI; and use AI as a tool online at home, and in the workplace'. Using the EU AI Act and the EU AI Literacy under Digital Strategy (European Commission, 2025) as additional sources, five learning outcomes were created focusing on informed use of AI, ethical considerations, and basic technical and practical skills at home and at the library. Once I had written the basic learning outcomes, AI (Anthropic, 2025) was used to help map them to Bloom's Taxonomy and extend them using the SMART model. Many points had been addressed, but AI helped me identify areas that needed further development, making suggestions which I considered for a final edit. This process reinforced the necessity of keeping my own critical thinking at the forefront of any AI implementation, using it as a tool rather than a replacement for pedagogical judgment.
During the design stage, work on the deliverables began by dividing the learning outcomes into topics and lessons and plotting them into a timeline. At this point, considerations and limitations - such as having no budget (the project was undertaken on a voluntary basis), time constraints, and most importantly, technological barriers - had the greatest impact. I realised I had to work with the tools the library has access to (e.g., Google Classroom and Canva Pro for non-profits), which were familiar to all volunteers, and find free or low-cost tools for the design. As I had used Genially before, and it has integration features with both Google Classroom and Canva, it proved to be the best choice for the learning resources. Additionally, with limited technological knowledge of LMS (learning management systems) and other more advanced tools, Genially presented itself as the most appropriate solution for the project.
I understand that I could have explored the budget further, digging deeper into hypothetical costs. However, with limited resources, I felt that my time and efforts would be better spent on areas that had a clear practical impact on my project.
As the whole course had been mapped at this stage, the second part of the project was on to the development stage, this was the most challenging process. While I am familiar with learning theories and have designed numerous eLearning resources using Genially and PowerPoint, for example, I had never designed the documentation behind those resources.
I am an experienced teacher who knows what effective teaching and learning looks like, but who had never given much thought to the design process behind it. I have been using and creating learning materials for so long, I had forgotten to look at how they were constructed. Designing a wireframe, screen plan, and storyboard were processes I had never undertaken before. I did not know where to start.
So I started with what I knew: I created the basic layout of an online lesson, a structured simple lesson plan, mapped to a lesson framework I am familiar with - engage, study, activate. According to Harmer (1998), the 'engage' stage gets learners interested by involving their emotions. The 'study' stage gets learners to focus on the content or information, while the 'activate' stage gets them to use the information in a real-life context. I also considered elements of Behaviourism, Cognitivism, and Constructivism. This blended theoretical foundation was a deliberate choice to meet the diverse needs of my learners. For instance, Behaviourist elements like repetition and low-stakes quizzes provided scaffolding and immediate feedback for 'Patricia,' who had little eLearning experience. In contrast, Constructivist approaches, such as problem-solving activities, offered opportunities for 'Aanya' to apply her existing knowledge.
With a clear outline using this lesson framework, I examined Gagne's Nine Events of Learning and realised that most had already been covered. All I had to do was include some other steps in the design, considering some of the missing Events, and my wireframe was completed. This approach - using my pedagogical knowledge to frame the learning experience - worked well as the foundations were clear to me, so I was able to build on what I already knew, from wireframe to screen plan to final storyboard.
I used Claude (Anthropic, 2025) to break down lesson points into bullet points. The use of Claude for content breakdown revealed both the potential and limitations of AI in instructional design. While it provided efficient initial structuring, I found that the AI-generated content required significant pedagogical refinement to ensure alignment with adult learning principles. This experience highlighted the importance of maintaining human expertise in educational decision-making while leveraging AI for efficiency gains. So instead of relying on AI to identify resources, I decided to do this myself. I researched and collated information, creating simple text for the screen plan, ensuring the content was accessible and used easy-to-understand terminology, avoiding cognitive overload. I added examples that were relevant to the learners, contextualising learning and planning for interactive and visual appeal.
The creation of the prototype was straightforward once all the information had been documented. Most importantly, there was still room for flexibility to ensure that design serves the learners' needs and consideration for Universal Design for Learning elements such as offering engagement through lesson hooks that could trigger imagination for both personas, options for social learning through forums in each lesson, external videos and downloadable resources in English and German, and assessment and information presented in a variety of formats, amongst other considerations.
The prototype also considered icons and images that were accessible and inclusive, sans serif font choice (Roboto), contrasting colours, and repetition of elements to create a sense of cohesion and consistency. Information was broken down into smaller chunks and presented in bullet points where possible, and opportunities to extend learning were provided in short videos. While many of these considerations were intrinsically part of my teaching practice, going through them systematically was an eye-opener, changing some of my ideas about what constitutes an effective learning resource.