AI simulation is changing what’s possible in workplace learning
- Joanna Smith
- 1 day ago
- 4 min read
Updated: 21 hours ago
Not long ago, if you wanted to create a genuinely immersive simulation in eLearning, you had to work hard for it.
I remember working with a highly experienced colleague on a Storyline course built around branching scenarios. She took scenario design seriously, and she was (still is!) a fantastic story writer. She wanted learners to feel the consequences of their choices as realistically as possible, so she designed the scenario with first-person video throughout. That meant writing scripts for multiple pathways, filming lots of short scenes, paying a group of actors, and then stitching everything together so the right scene appeared at exactly the right moment depending on what the learner chose.
It was excellent learning design. For the time (and even now!) it was brilliant.
It also took two full days of filming, a large amount of editing, and a lot of development time to bring it all together. And once learners had been through the scenarios once or twice, they had more or less seen the range of what was possible. The simulation was finite.
That is why AI simulation feels like such a significant shift.
The pedagogy is not new. Simulation has been a strong learning approach for a long time. What is new is that we can now create rich, realistic practice experiences much more quickly and affordably than we could before.
That matters for any organisation trying to help people build confidence in difficult workplace conversations.
We have seen this first-hand in our own work. In one project, we designed AI role-play modules for leaders so they could practise common but high-stakes conversations such as giving praise, coaching someone to improve, and having difficult conversations. In another, we created simulations for frontline, community-facing support workers. For that work, we built AI personas with their own personalities and back stories, so learners could work through realistic support conversations rather than generic prompts.
That is where this gets interesting. The learner is not just clicking through content or choosing from a fixed set of pre-written responses. They are practising. They are responding in the moment. They are trying out language, adjusting their approach, and seeing how the interaction unfolds. And they are getting personalised feedback.
And because the exchange is generated fresh each time, they can do it again and again without simply repeating the same script.
With older branching courses, even excellent ones, there was always a practical ceiling. You could only build so many paths. You could only film so many scenes. You could only justify so much production cost. With AI simulation, the cost structure changes. You are no longer paying for two full days of filming, multiple actors, and hours of editing just to create the sense of a real conversation.
Instead, you can invest in the thing that matters most: designing a strong learning experience.
That does not mean AI makes the work easy. It just shifts where the effort goes.
We still need to think carefully about what the learner is practising. We still need realistic contexts. We still need believable characters. We still need clear criteria for what good performance looks like. We still need testing, a lot of testing. We need to know whether the simulation is behaving in useful ways, whether the prompts are leading learners in the right direction, and whether the feedback supports the learning goal.
AI does not remove the need for good instructional design. If anything, it makes that need more obvious. A simulation can look impressive very quickly, but that does not automatically make it useful. If the conversation is unrealistic, if the challenge is too vague, or if the learning goal is muddled, the experience will still fall flat.
In other words, a fool with a tool is still a fool.
That is one reason I find the current conversation about AI in learning slightly frustrating. Too much of it focuses on speed alone. Speed matters, of course. Cost matters too. But the bigger point is that AI gives us a way to make strong pedagogy more accessible.
And simulation is strong pedagogy.
That is backed up in the research as well. Sellberg and Lindwall (2026) make the point that simulation is not just a place for rehearsing isolated skills. It is a setting where professional knowledge is shaped and reworked through participation. Their line that really stands out is this:
“Across the studies, simulation emerges not merely as a tool for skill rehearsal but as a setting where professional knowledge itself is produced, debated, and reconfigured.”
That captures why simulation matters so much in workplace learning. It gives people room to practise in context, make decisions, respond to complexity, and build confidence in situations that actually resemble the work.
For years, simulations of this quality were possible, but only if a project had the budget and time to support them. Now they are becoming a realistic option for many more organisations. Not because the design work has disappeared, but because the production model has changed.
At the time of writing, we are mainly working with text-based interactions, and they are already extremely useful. Learners can step into realistic conversations, test different approaches, and get more practice than they ever could from a branching scenario. We are also experimenting with voice-to-voice interactions, which opens up another layer of realism. That is still emerging, but it is clearly where things are going.
If your current eLearning is largely content-heavy and low on practice, AI simulation gives you a new way to close that gap. It allows you to create scenario-based practice that feels more alive, more responsive, and more relevant to real work. And it no longer has to cost an arm and a leg to get there.
Give it a try
The best way to understand this is to try it. Click the link below to have a go with an AI simulation.
