The future of learning: why competency-based training and AI role-playing finally tackle the transfer problem
Most organisations already know this. You send someone on a training course, they come back full of good intentions, and six weeks later, little of what they learnt that day remains. Not because the trainer was bad, not because the employee wasn’t willing. But because one-off learning in a controlled environment simply doesn’t translate into different behaviour on the work floor. That is the transfer problem, and it has existed for as long as organised training has.
Competency-based training changes the rules of the game. Not by intensifying training, but by organising it fundamentally differently: focused on demonstrable behavioural change, repeatedly practised in realistic scenarios, and measurable down to the competency level. And with the rise of AI role-playing, this is becoming scalable for the first time.
The transfer problem: an issue the sector has faced for decades
In 1988, Timothy Baldwin and Kevin Ford wrote an influential article in the journal Personnel Psychology that would later become one of the most cited studies in training literature. Their conclusion was uncomfortable: only a small fraction of what people learn during training is actually applied in the workplace. More recent analyses estimate that figure at less than ten per cent of training content that systematically translates into different behaviour in the workplace.
That is no small problem. It is a systemic failure in how we organise learning. And yet, in many organisations, the model has hardly changed: a training programme, a module, sometimes some e-learning afterwards. The context of learning and the context of application remain fundamentally separate.
What makes transfer so difficult? Research repeatedly points to the same factors: a lack of opportunities to practise after the training, a working environment that does not support the new behaviour, and the fact that learning in a safe setting bears little resemblance to the chaos of real-world practice. People learn something in a classroom, but have to apply it at a desk, in a meeting, at a time when they also have ten other things on their minds.
What makes competency-based training different
Competency-based training starts with a different question. Not: ‘What do people need to know?’, but: ‘What do people need to be able to do, and how do you recognise that they can do it?’ That shift sounds subtle, but it has far-reaching consequences for how you design training.
In a competency-based model, learning is not measured by attendance or passing a test. It is measured by demonstrable behaviour in context. This means that, as an organisation, you must determine in advance which competencies you wish to develop, how you will make those competencies observable, and what evidence you will accept as proof of mastery. PractAIce is built on precisely this foundation: every practice session is linked to specific competencies, and progress is made visible at the behavioural level, not based on attendance or self-reporting.
For soft skills, such as communication skills, giving feedback, dealing with resistance and conflict management, this is no simple task. These skills are, by definition, contextual. How someone gives feedback to a junior colleague is different from how that same person gives feedback to a defensive director. Competency-based training recognises that complexity, rather than simplifying it. PractAIce translates that recognition into practice: for each competency, scenarios have been developed that challenge precisely that nuance.
Role-play as a learning tool: from awkward exercise to serious instrument
Role-plays have a mixed reputation in the training world. People find them awkward, forced, or experience them as a kind of theatre that has little to do with their daily work. That experience is understandable, but says more about how role-plays are used than about their potential.
The science is clear on the importance of simulated practice. In his research into the development of expertise, psychologist K. Anders Ericsson demonstrated that exceptional performance does not stem from talent or years of experience in themselves, but from deliberate practice: targeted, repeated practice on specific weaknesses, with immediate feedback. That is precisely what a well-designed role-play does.
A study in the journal of the National Institutes of Health showed that scenario-based role-playing exercises for training soft skills, in this case among medical students, led to significant improvements in communication skills across successive practice sessions. What stood out was that students explicitly stated that they learnt through role-play what remained unattainable through lectures. The exposure to realistic situations made all the difference.
The problem with traditional role-playing is scalability and safety. An exercise with a colleague rarely feels real, because you know the other person is playing along. And in a group session, nobody wants to make a fool of themselves in front of the group. That barrier hinders the quality of learning.
Customised role-playing: why customisation is key
Not every role-play is the same. That may sound obvious, but its implications are often underestimated in training practice.
A generic scenario about ‘conducting difficult conversations’ has a different learning value than a scenario built around the specific situations people encounter in their own roles. A team leader in the healthcare sector has different conversational contexts than an account manager in the B2B sector, or a manager at a government agency. The degree to which the scenario mirrors one’s own reality partly determines how seriously the brain takes the situation, and thus how much is actually learnt.
Research into training evaluation shows that the degree of similarity between the training context and the application context is one of the strongest predictors of successful transfer. Baldwin and Ford describe this in their transfer model as one of the central mechanisms: the greater the similarity between the learning and application situations, the greater the likelihood of transfer.
Customised role-plays build in that similarity. PractAIce allows organisations to create scenarios that align with their own team dynamics, culture, typical conflicts and the competencies the organisation wishes to develop. The AI avatar plays a recognisable role, not a generic situation that seems to come from another planet.
How AI tackles the transfer problem in a fundamentally different way
What AI adds to competency-based training is something the sector has long been missing: the ability to organise repetition without relying on a trainer, a colleague or a schedule.
Transfer so often fails because there is no further structured practice opportunity after the initial training. Knowledge and skills fade. The work environment reverts to familiar behaviour. Recent literature on learning and transfer processes in organisations emphasises that transfer is not a single moment, but a process that requires support long after the training. AI makes that support possible on an ongoing basis.
An employee practising via PractAIce can start a scenario at any time. Five minutes before a difficult conversation, or a week after something went wrong in a meeting. The exercise aligns with the moment of need, which is precisely the circumstance under which the transfer of training proves to be most effective.
Furthermore, AI makes it possible to measure competencies objectively across multiple training sessions. That is the promise of genuine competency-based training: not simply ticking a box, but demonstrating that someone has mastered a skill. After each role-play, PractAIce generates detailed feedback on specific behavioural indicators — does someone listen actively, does someone structure feedback in a way that the other person can hear, does someone remain calm when emotions run high — and thus makes visible what would otherwise remain invisible.
Frequently asked questions about competency-based training and AI role-plays
What is the difference between competency-based training and traditional training?
Traditional training is usually focused on knowledge transfer: what does someone need to know or understand? Competency-based training focuses on demonstrable behaviour: what does someone actually need to be able to do in their own work situation, and how do you demonstrate that? The difference lies in the yardstick. In competency-based learning, the standard is not a completed module, but proven behavioural mastery in context.
Is competency-based training also suitable for soft skills? The competency-based model is particularly well-suited to soft skills. Communication skills, giving feedback, and dealing with resistance are skills that cannot be assessed in a test. They can only be assessed in practice, or in a realistic simulation of it. Competency-based training provides the framework; AI role-plays provide the practice environment.
How does an AI role-play differ from a traditional role-play?
The biggest difference is safety and repetition. In an AI role-play, there is no audience, no colleague playing the ‘game’ alongside you, and no embarrassment when mistakes are made. Furthermore, you can run through the same scenario multiple times, until the desired behaviour becomes second nature. This is precisely the structure of deliberate practice that research has shown to be most effective for behavioural change.
What does it cost to implement PractAIce for a team?
PractAIce operates on a licensing model tailored to the size of the organisation and the number of intended users. There are no travel costs, venue costs or reliance on external trainers. For a specific quote, we recommend getting in touch or requesting a demo.
Finally: has the future of learning already begun?
The transfer problem has existed for decades. But the solution simply wasn’t there yet: there was no scalable way to let people practise sufficiently, in realistic situations, with immediate feedback, measured at a competency level. That has now changed.
Competency-based training using AI-powered role-plays is not just a pipe dream. It is already being used by organisations that understand that learning is not a one-off event, but an ongoing process. Research by LinkedIn shows that 91% of L&D professionals believe continuous learning is more important than ever. But continuous learning requires an infrastructure that makes it possible. PractAIce provides that infrastructure.
Are you unsure whether this aligns with your organisation’s learning ambitions? A demo provides more insight than a brochure ever could. Feel free to request a demo. No obligations – just see if it’s a good fit.