Practical strategies for integrating AI into eLearning
Integrating AI into e-learning can solve real challenges, but starting without a clear purpose often leads to wasted effort. AI is powerful, but like any tool, its effectiveness is determined by how well it fits your specific learning goals. Too many organizations fall into the trap of chasing buzzwords and launching flashy pilots that never address real-world needs.
AI generated image / dominknow.com
In a recent episode of the podcast IDIODC (Instructional Designers in the Office Drinking Coffee), Tim Hundley, Chief Product and Technology Officer at Welbee, shared practical insights on meaningfully integrating AI into training experiences. Welbee, a UK-based company focused on improving the wellbeing of educational institutions, has successfully implemented AI across the curriculum. Their work provides valuable lessons for training teams looking to improve content development, learner engagement, and measurable impact.
Start with real problems, not technology
Tim emphasized avoiding the trap of implementing AI just because it’s trendy or required by leaders. “We wanted to make sure that we not only incorporated AI in a meaningful way, but that it wasn’t just a buzzword,” he said.
The most successful implementations start with clarity. Instead of asking, “How do we leverage AI?” ask, “What are the specific problems we are struggling to solve?”
The main issues with Welbee are:
It allows users to quickly navigate through a vast array of welfare resources. We provide 24-hour support across time zones around the world.
These challenges led to our first AI application: a chatbot trained on course content. The chatbot acted as an always-available support system, helping learners find the right information without the need for human intervention.
This approach reflects a broader industry trend toward implementing AI through targeted pilot programs that address real needs. Smart instructional designers use AI not as a replacement for expertise, but as a co-pilot to speed up content creation and tackle the “blank page problem.”
Build on success
Once Welbee proved the value of chatbots, it expanded its AI integration into other areas of e-learning. Currently, AI-enhanced experiences include:
Course-specific chatbots: Trained on relevant content for targeted answers. Interactive knowledge checks: Provide instant, contextual feedback instead of generic responses. Dynamic role-playing scenarios: Allow learners to practice coaching conversations in a safe simulated environment. Personalized learning paths: Suggest custom routes based on learner progress, goals, and areas of difficulty.
One notable example is coaching exercises. Originally provided as a static handout, this activity intimidated learners and was underutilized. By incorporating AI, Welbee transformed them into interactive role-play sessions. The chatbot acts as a teacher seeking guidance and the learner practices multiple conversations. The system provides structured feedback based on a coaching model and strengthens skills through safe repetition.
This shift highlights an important lesson: AI does more than just automate tasks. You can reimagine the learning experience in a way that’s more engaging, personalized, and impactful.
Important technical considerations
Behind the scenes, several technical decisions shaped Welbee’s success.
Content management: AI should be based on carefully selected course-specific materials to ensure accuracy and relevance. This reduces the risk of “hallucinations” and inappropriate reactions. Temperature settings: Temperature affects creativity. Less reliable, more diverse, but more risky. Welbee lowered this value in most scenarios to ensure that the response complied with the provided content. The setting range is 0 to 2, with lower values producing more predictable output.
AI generated image / dominknow.com
Prompt engineering: Creating clear prompts is now a core skill for instructional designers. At Welbee, prompts defined the persona, tone, context, and boundaries of the AI. For example: Persona: “You are an expert instructor specializing in leadership skills using the GROW model.” Context: “Use British English and British educational terminology. See course content.” Rules: “Use British English, focus on coaching, and are always collaborative.” Quality Assurance: All AI features are rigorously tested, including edge cases and sensitive topics. Creative scenarios required deeper validation to avoid mistakes.
These details may seem technical, but they will determine whether your learners have a smooth and reliable experience or a frustrating experience.
Managing bias and ensuring safety
AI can unintentionally perpetuate hidden biases in training data, engine design, and even prompting rules. Welbee mitigated this by:
Limit AI to vetted course-specific content. Use low temperature settings for predictable output. Build guardrails with precise prompts. Perform regular monitoring and updates. Clearly label AI features such as the “Nebula AI” chatbot.
Sensitive subjects such as mental health required additional safeguards. For example, health conversations automatically include hotline numbers and links to official resources. Transparency was essential. When learners know they are interacting with an AI, they feel more confident and secure.
As the Brookings Institution points out, bias can manifest at three levels: the data, the engine, and the human rules that are applied. [1] Addressing all three is essential for fair and ethical adoption of AI.
Measuring real impact
Adding AI for novelty is of little value. The real test is whether learners are more engaged, successful, and supported. Welbee tracked specific results, including:
Higher course completion rates. Increases engagement compared to passive modules. Positive feedback from learners, especially regarding chatbot support. You will spend more time exploring resources. Retain important concepts more strongly.
Imagine this. Learners don’t just finish the course, they stay longer, practice more, and report better results. That’s a meaningful impact.
It’s not just about wellbeing. Companies like IBM, Amazon, and Walmart are reporting similar results when AI supports personalization, real-time feedback, and adaptive learning. [2]
Practical implementation guidance
Tim provided clear guidance for teams considering AI.
Start small: Identify one problem that AI can solve today. Be transparent: Communicate to learners when and why AI will be used. Enhance rather than replace: Put human expertise front and center. Frequency limit: Aim for 3-4 AI touchpoints per 2-hour course. Test thoroughly: Cover edge cases and stress test sensitive topics.
This disciplined approach prevents the AI from overwhelming the learner or becoming a gimmick.
way forward
As AI evolves, so will e-learning opportunities. Welbee’s story shows that the most effective applications are purpose-driven, carefully controlled, and focused on the learner.
Instructional designers need to see AI not as a shortcut, but as a partner in creating richer experiences. New possibilities include:
AI-driven emotional intelligence tools that adjust responses based on learner emotions. Immersive practice environment with AR/VR integration. Predictive analytics warns you of the risk of learner disengagement.
For L&D teams, the message is clear. The implementation of thoughtful AI enhances the effectiveness of training and positions learning as a driver of organizational success.
Build a better learning experience
The foundation of a great learning organization is the systems, tools, and teams that enhance the learning curriculum and drive continuous improvement. It’s not just about delivering courses, it’s about building sustainable structures for collaboration, efficiency and innovation across the learning lifecycle.
Dominnow | ONE provides an all-in-one LCMS and authoring solution that brings these elements together. By enabling instructional designers, subject matter experts, and learning leaders to collaborate seamlessly, dominKnow | ONE ensures that creativity and expertise is shared rather than siloed. This foundation, combined with thoughtful integration of AI, enables organizations to create engaging, accessible, and future-ready learning experiences that scale with their evolving needs.
References:
[1] Artificial intelligence and bias: four key challenges
[2] Case study: Successful implementation of AI in corporate training
Read more: First published on: www.dominknow.com
Source link
