An interview with Thomas Lührs
Role-play scenarios have been a staple of leadership development programs for decades. Yet many participants would probably prefer to skip it altogether. Does it have to be that way? Anyone who has attended a communication skills training session knows the scenario: After the theory comes the practice. Two participants are asked to come to the front of the room. One plays the manager, the other a difficult employee. Their colleagues watch. For some, this works extremely well. Others have just one thought: Please don't pick me.
That's exactly the challenge we discussed with Thomas Lührs. He works in project management for digital initiatives and new training programs at the maritime competenzcentrum, where he explores how innovative digital learning formats can be integrated into communication skills training. The organization has already tested AI-powered conversation simulations from 3spin Learning with frontline employees and aspiring leaders. One of these approaches involves AI-powered role-playing, where employees practice realistic workplace conversations with an AI avatar.
In this interview, Thomas explains how AI avatars can complement traditional role-play exercises, why repetition is essential for transferring learning to the workplace, and where human coaches still have an advantage over AI.
Thomas Lührs: It really depends on the individual. Some people find it easy to step into a role. But there are also people who break into a sweat at the mere mention of role-play. They may not feel comfortable opening up in that situation, or they simply don't want to. Perhaps they're worried about embarrassing themselves in front of their training partner or the rest of the group. And that naturally affects the exercise. After all, the goal is to practice a conversation that feels as realistic as possible. But if I'm also worrying about how my colleague perceives me, that creates an additional barrier.
Thomas Lührs: For me, the biggest difference is that it creates a private practice environment. I'm talking to an avatar that is initially completely neutral, and I have no relationship with it outside the training situation. I can do this on a computer or, for example, using a VR headset. There's still someone standing in front of me, and I'm still having a conversation. But it feels very different from practicing with a colleague in a training session. That's something we definitely recognized as an advantage during our testing.
Thomas Lührs: Yes, absolutely, in certain situations. Especially when training difficult conversations, the whole point is to experiment with different approaches. If I'm constantly worried about embarrassing myself in front of someone else, I'm probably going to be less willing to experiment. With an avatar, on the other hand, I can try something completely different and see what happens. And afterward, I receive feedback on my approach.
Thomas Lührs: The level of acceptance was very high. As part of a research project, we tested these applications with different groups, including frontline employees and people preparing for or developing further in leadership roles. The feedback was very positive. Participants also recognized the learning benefits and saw the value of this type of training. Two things are particularly important to me:
Why is repetition so important? Because it allows me to actually apply the feedback I receive. After a simulation, for example, I might get a suggestion for improvement. I can then repeat the same conversation, or a similar scenario, and consciously try to implement that suggestion. Afterward, I can see how my change in approach affected the conversation. I can even deliberately try something unusual just to see what happens. This kind of repeated experimentation is relatively easy with an AI avatar.
Thomas Lührs: I see it more as a complement. For example, AI role-play can be integrated directly into a training session. You start with the theory and then move into practical application. Instead of having participants practice a traditional role-play with another person, they can complete that part of the training with an AI avatar. There's a second opportunity that I find just as exciting: what happens after the training. Participants can independently revisit and practice what they learned in the workshop, helping reinforce those skills over time. This creates a blended learning approach that connects instructor-led training, digital practice opportunities, and the application of new skills in everyday work.
Thomas Lührs: That's certainly one of the biggest opportunities. If I training something just once during a workshop, I haven't developed a routine or built lasting confidence yet. With digital role-play, I can practice whenever I need to. Maybe I know I have a difficult conversation coming up in an hour. I can take a few minutes beforehand to practice that exact situation with an avatar. If I just stand in front of a mirror and rehearse the conversation, I don't get any reaction. Avatars respond to what I say, and afterward, I receive feedback.
Thomas Lührs: That's a difficult question. Human coaches still have advantages today. AI initially operates based on how it has been configured and prompted. A human trainer can respond more flexibly to unexpected situations. There are also factors such as emotions, gestures, facial expressions, and the subtle nuances of a conversation. In some cases, humans are still better at recognizing these signals and interpreting them in context. That's why I don't necessarily see it as an either-or decision.
Thomas Lührs: Above all, in its availability and the ability to practice repeatedly. A human coach can't be by my side whenever I need one. An AI-powered learning companion, on the other hand, could be available at exactly the right moment. To me, this moves us toward lifelong learning and ongoing development that's integrated directly into everyday work. And in the long term, we could take that idea even further.
Thomas Lührs: Some kind of digital learning buddy could be an interesting development. If that assistant understands my context, it could support me more effectively throughout my workday and provide feedback afterward on a presentation or a conversation. But for that to work, the AI would need enough information about me to offer more than just generic recommendations. And that's where we quickly run into an important limitation.
Thomas Lührs: Absolutely. As soon as AI starts supporting or recording conversations, several questions immediately arise:
These issues can also influence how openly people communicate in the first place. Even in human coaching, I need to trust that I'm in a confidential, safe environment. When conversations are recorded and analyzed digitally, that question becomes even more important. It's something we absolutely need to consider when thinking about these future scenarios.
Thomas Lührs: In my view, organizations shouldn't simply introduce AI everywhere just because the technology makes it possible. First, they need to identify which processes and use cases would actually benefit from AI. If I take a poor manual process and simply digitize it or add AI, I may end up with a poor AI-powered process. That doesn't automatically make it better. This is also why leaders need to learn how to evaluate AI appropriately and provide clear guidance to their teams.
Our conversation with Thomas Lührs highlights an aspect of AI in learning and development that receives relatively little attention. Sometimes, the greatest advantage of an AI avatar may be precisely that it isn't a coworker. When it comes to soft skills development, consuming information isn't enough. Communication, giving feedback, and managing conflict are skills that require hands-on practice.
Traditional role-play provides that opportunity, but it also introduces social pressure that can hold some learners back. AI-powered role-play changes the dynamic. Employees can experiment, make mistakes, receive feedback, and try again, without a colleague from their own team watching and evaluating their first attempt. This is the principle behind AI-powered soft skills training. Train real-lifec workplace conversations, receive personalized feedback, and immediately apply what you've learned in another simulation. That doesn't automatically make AI a better coach, nor does it make human trainers obsolete. But it creates an additional opportunity for practice between learning a concept and applying it in the real world.
That's exactly where 3spin Learning's AI-powered soft skills training comes in. Organizations can create realistic workplace conversation scenarios for employees and leaders to train with AI-powered role-play. The key difference compared with traditional e-learning is that learners have to actively participate in the conversation. They must speak, respond, and make decisions rather than simply consume content. They experience realistic responses, receive feedback, and can repeat scenarios until theoretical knowledge gradually turns into practical confidence.
This brings together several elements that are particularly valuable in soft skills development: