By Kevin Shepherdson, Founder & CEO, Straits Interactive
“When you’re lost, there's always “Please Suggest” for a start ” - Reflections from a course participant on using Generative AI to inspire initial content drafts for refinement after.
In April 2025, a group of forward-thinking educators, trainers, and learning and development (L&D) professionals came together for our hands-on course on Generative AI Apps Design and Prompt Engineering in Learning and Development, conducted with Singapore Management University (SMU) Academy. They walked away with more than just hands-on experimentation in developing Generative AI apps for lesson preparation and learning supplementation - they also acquired a deeper understanding of the principles required to deploy such technologies responsibly in education.
Anchored by the ISO/IEC 5338 AI System Development Life Cycle, participants learned to design and deploy three simple, yet functional Generative AI tools:
1. Lesson Builder for assisting tailored content creation
2. AI Tutor for guided self-paced learning, trained on publicly available syllabi from Singapore’s Ministry of Education (MOE)
3. AI MCQ Assessor for automated quiz generation
The true breakthrough, however, wasn’t just in building apps. It was the understanding of what responsible applications of Generative AI should look like in learning environments. Naturally, making sure the apps were reliable was not as straightforward as the end products appeared.
Behind the Bot: Building with Purpose
Participants were guided through what one called the “sausage factory-like process of app genesis to the final outcome.” In other words, they learned not just how to write and feed prompts to Large Language Models (LLMs), but how to:
1. Align Generative AI tools to learning objectives
2. Follow structured app development using the ISO 5338 framework
3. Integrate continuous validation and governance checks into the app development process
Stages of the ISO/IEC 5338 AI System Development Lifecycle Model
This wasn’t just a glorified form of prompt hacking. It was a practical, applicable form of responsible Generative AI app development, designed with learner safety, accuracy, and maintaining educator control in mind.
When AI Gets it Wrong - and What We Did About it
During the course, participants used the local MOE syllabus to contextualise their AI Tutors, with promising results in delivering personalised lesson support.
However, generating Multiple-Choice Questions via the AI MCQ Assessor to test learners’ understanding revealed a common and well-documented flaw in Generative AI: hallucinations. Some questions were factually incorrect or poorly phrased.
Instead of simply working around this, we took it as a teachable moment. We introduced test sets to evaluate question quality along with strategies in bias detection, prompt tuning, and grounding techniques to strengthen the accuracy of the AI MCQ Assessor’s outputs.
The key takeaway: Treat AI like a junior assistant - helpful, but still learning the ropes.
Comments from the Classroom
The 2-day session illuminated the inner workings of Generative AI for educators in the room, who gained new clarity on what it takes to navigate Generative AI safely by taking the best available measures to circumvent its insidious parts.
Some found it “fascinating to explore the behind-the-scenes process of chatbot development,” while others appreciated learning how the ISO 5338 provides a framework for responsible AI deployment in education. One mentioned that she would “always remember the mitochondria…and the capital of France,” as a memory aid for how to verify AI-generated content, like that of the AI MCQ Assessor, and the importance of correct data inputs in prompts.
Irrespective of background, be they educators, trainers, or tech-savvy learners, all attendees left with practical tools and a stronger foundation in AI literacy.
The most poignant reflections, however, came not from the course content but from career experience. Sharing an anecdote about a TikTok employee who was retrenched from a content moderation role after failing to upgrade his skills, a participant expressed a sobering reminder of the criticality of continuous reskilling and upskilling as working adults in a changing world.
Moving from Exploration to Application
A future-ready mindset is essential across all sectors - and cultivating this is central to the courses we offer at Straits Interactive. But it’s not about replacing teachers with Generative AI. It’s about sharpening educators with tools that make differentiated learning scalable and safe.
Whether you work in schools or corporate training, the imperative is clear:
1. Design responsibly
2. Deploy ethically
3. Reskill continuously
For educators, consultants, and L&D leaders seeking to hone their technological competency and understanding of Generative AI, opportunities are readily available. Our next course intake is open, with more to follow. No programming background is required - only a willingness to learn, build, and adapt before the tide of change outpaces preparation.
If you are interested in adopting generative AI-powered EdTech safely and securely, our Capabara platform is currently available on a limited trial basis. To work with us, write in to sales@straitsinteractive.com with your organisation’s email address. You can also stay tuned to Capabara’s latest developments by following CAPABARA on LinkedIn or heading over to capabara.com to find out more.
This article was originally published on 29 April, 2025 at the Governance Age
