Featured in Learning Guild

While Generative AI is often viewed through the lens of automation and productivity, its real value in learning may lie elsewhere—in surfacing what we never thought to look for.

When integrated systematically into corporate training environments, GenAI goes beyond streamlining tasks: It becomes a mirror that reflects behaviors that are often overlooked, needs that are unmet and questions that were waiting to be raised.

In one such experiment, a GenAI-powered learning assistant was embedded into the flow of corporate learning, serving as a round-the-clock digital coach, answering questions and helping learners access content. The aim was to offer real-time support, answer learner queries and help the learners navigate training materials more effectively.

The results exceeded our expectation—the outcome was about more than just enhanced accessibility. Serving as a diagnostic lens, the AI tool had inadvertently revealed unexpected gaps and behavioral preferences of the learners.

Its actual value lay not only in delivering information but also in revealing questions the learners wanted to ask.

Key insights

While numerous insights emerged, several stood out and challenged assumptions—helping us see learners in a different light:

• 63% of queries focused on edge cases or exceptions to rules, not the standard core procedures that were documented over several months. Our learners weren’t confused about the “what”—they needed clarity on the “what if.”
• Drop-offs spiked at PDF links. More than 80% of users who clicked on a resource didn’t make it past the first page. The message was clear: no one wants to engage with a 14-page document when they’re looking for a quick answer during their workday.
• Structure & searchability go hand-in-hand. One of the most comprehensive modules received near-zero queries from learners—not because of how perfectly it was structured, but because it was almost impossible to search. The AI bot revealed that learners were using terms that did not align with the registered taxonomy. In essence, it was a well-designed but underused course.
• Learners still asked basic navigation questions like “where do I click?” Despite our assumptions of digital maturity, the bot highlighted a huge disconnect between the information architecture and the actual user experience. This meant we must meet learners where they are—not where we assume they should be.

Our response

In response to the above insights, our learning team came up with a revamped approach toward learning content. We decided to:

• Address the “gray zones” that learners kept asking about. This was possible by restructuring content into micro-modules that focused only on edge cases.
• Embed video snippets directly into the bot’s responses, bypassing full courses entirely for high-frequency questions.
• Re-tag the content library based on the natural language drawn from actual learner interactions, instead of internal jargon.

Design for discovery

Our biggest realization was that  an AI assistant could integrate and evolve into a listening tool with a steady feedback mechanism. It could fluently highlight learner struggles, system disruptions, and complicated curves in learning design. The value of our chatbot is not just in the answers it gives—it is in the questions it surfaces.

Too often in L&D, we design for delivery. However, GenAI offers us a new opportunity: to design for discovery.

The real promise of AI in learning isn’t automation. It’s illumination.