Becoming a teacher requires more than subject expertise: it demands the ability to navigate emotionally charged and challenging interaction situations with professional judgement, resilience and sensitivity. Research indicates that teachers’ emotion regulation is associated with the quality of teacher-student interactions and with indicators of teaching effectiveness (Aldrup et al., 2024). In vocational education, teachers frequently engage with diverse learner groups, which can make interactional situations more varied and, at times, emotionally demanding. For novice teachers in particular, challenging classroom situations can be experienced as stressful and professionally demanding. Teacher students begin to encounter these demands during practicum and early teaching experiences. Teaching in authentic classroom contexts has been conceptualized as requiring advanced social and emotional competencies, including the ability to regulate one’s emotions in interactional settings (Jennings & Greenberg, 2009). Emotion regulation is commonly defined as the processes through which individuals influence the emotions they experience, when they experience them, and how these emotions are expressed (Gross, 1998).
Digital transformation and the emergence of AI offer new possibilities for creating learning environments that support the development of these complex professional competencies and build confidence in teaching. While technologies such as virtual reality (VR) can offer safe and repeatable opportunities to rehearse challenging professional scenarios, international policy frameworks emphasize that such tools should be integrated in pedagogically and ethically informed ways that support – rather than displace – teachers’ professional judgement and expertise (UNESCO, 2023). From a professional learning perspective, simulations can provide structured spaces for reflection-on-action, a concept introduced by Schön (1983). Research on simulation-based teacher education further suggests that structured scenarios combined with guided reflection can support the development of professional judgement and perceived preparedness by allowing novice teachers to rehearse decisions in controlled settings (Badiee & Kaufman, 2015; Chen, 2022). This perspective is particularly relevant in vocational teacher education, where professional identity often develops at the intersection of prior occupational expertise and emerging pedagogical roles (Beijaard et al., 2004; see also Köpsén, 2014).
This pilot examined how VR- and AI-based simulations can be used to create reflective learning experiences for teacher students when encountering challenging interpersonal situations, particularly in relation to emotional demands, professional judgement, and professional identity. The goal was to provide a safe and repeatable environment where teacher students could practice de-escalation, experiment with different strategies without real consequences, and reflect on the question: Who am I as a teacher in a difficult situation?
This text describes how the pilot was designed, what students experienced during the VR–AI interaction, and what educators learned about the potential and limitations of integrating AI-based simulations into teacher education. Through this lens, the experiment provides insight into how emerging technologies can be used to support reflective practice around resilience and interaction in classroom contexts, while emphasizing that technology should serve pedagogical aims rather than drive them.
The experiment: Challenging classroom situation de-escalation
The simulation took place during a learning session in an internationally oriented teacher education programme that grants the pedagogical qualification for teaching in Finland, including vocational education. Teacher students were informed about the opportunity to practice a challenging interaction scenario with a simulated vocational student, represented by an AI-driven virtual avatar in a VR environment. They were invited to participate on a voluntary basis. Participants were also asked for consent to publish anonymised reflections on their experiences in a Jamk publication. Six students volunteered and participated in the sessions in May 2025.
The simulation was designed to give teacher students a safe, repeatable opportunity to practice navigating an emotionally challenging interaction with a learner. The situation involved a teacher student and an AI-driven virtual avatar simulating a student in a VR environment. During the simulation, the student lost control in the classroom and threw an object after receiving exam results.
The VR-simulation functioned as a pedagogically grounded activity rather than merely a technological experiment, with a focus on reflective engagement around emotional resilience, professional judgement, and self-awareness.
The situation was constructed with following phases:
Phase 1 – Preparing for the simulation scenario and reflection
Those students who expressed interest received background material a few days beforehand, introducing the simulated setting, student profiles, and events leading up to the scenario. The material invited participants to consider who they are as teachers in such a situation and to reflect on the broader purpose of assessment. Practical guidance, including information on potential motion sickness in using a VR headset and how to manage it, was also provided.
Phase 2 – VR Simulation with an AI-Driven Virtual Avatar Representing a Simulated Student
The session was held on campus in two shared classrooms. Before entering the VR environment, the students received instructions on how to use the VR headsets and controls to interact with the simulated student. The teacher student communicated with the simulated student through spoken interaction. The simulated student responded using speech, written text (speech bubbles), and body language. After practising with the equipment alongside technical experts, teacher students reviewed the case instructions. The scenario was replicating a classroom in a vocational college, where teacher students assumed the role of the teacher and navigated a challenging situation with the simulated student in a heightened emotional state. The emotionally charged scenario required them to de-escalate the situation and to experiment with different strategies without real-world consequences. Teacher students began the simulation at their own pace, engaged with it for as long as they felt necessary, and were free to take breaks as needed. Two to three teacher students worked individually in the same room during the exercise, with technical support readily available, while teacher educators remained outside the simulation space to ensure an undisturbed experience.
Phase 3 – Identity reflection in debriefing
Following the simulation, teacher students reflected on their experiences of interacting with the simulated student, who they are as teachers in challenging situations, their pedagogical principles, and the purpose of assessment. The central question guiding the discussion was: “Who am I as a teacher in a challenging situation?” The guided reflection focused on this question as well as the feelings of the teacher students. A teacher educator facilitated the reflection and took notes. The discussion lasted approximately 30 minutes.
Technology serving pedagogy
The technical choices were made by the software development company Virtual Dawn. Particularly the modelling of emotional states and coherent behaviour were essential for creating a simulation that felt authentic enough to trigger meaningful reflection. This emotional realism was critical for supporting the identity-building aims of the pilot.
Using AI or VR in pedagogically meaningful ways requires new skills and, often, specialised services to ensure the technology supports educational goals. Achieving “pedagogical realism” calls for systems that can generate not only verbal responses, but also emotionally plausible behaviour – such as the reactions of a distressed learner – which basic AI characters rarely provide. More advanced platforms or targeted customisation, like three-dimensional models, AI, and behavioural modelling used in this simulation, are typically needed to create coherent, realistic interactions.
The Jamk solution was an early demonstration of this approach. Ongoing development is extending these capabilities toward more sophisticated Non-Player Characters (NPCs) that can manage emotional states, memory, and complex behavioural structures. In this context, the NPC represents a simulated student embodied as a virtual character whose responses and behaviors are generated in real time by an AI-powered conversational system to create interactive, pedagogically meaningful practice situations. While this technology is still in the early stages of development, continued refinement aims to deliver increasingly powerful tools for immersive learning and generative, LLM -based- AI interactions that meaningfully support pedagogical goals.
By enabling consistent emotional cue – such as frustration or perceived unfairness – the design allowed students to practice responses that resemble real classroom situations. Aligning AI-driven virtual avatar behavior with pedagogical intent ensures technology serves the learning objectives rather than overshadowing them, and this emotional realism is essential.
Teacher student outcomes
Teacher students reported that the VR environment enabled them to experiment with various ways to respond to a distressed learner and evoked emotions. While some found the simulated student repetitive or easy to resolve, others used the opportunity to test multiple interaction strategies. Some students engaged with the simulated student for an extended period to try out different approaches to navigating the situation, whereas others felt that the scenario offered a straightforward path to resolution and concluded it quickly.
A central outcome was that teacher students reflected more deeply on their emerging teacher identities than on the purpose of assessment which was one of the objectives of the training. They emphasised the importance of building trust, understanding learner needs, communication and solution-orientedness in challenging situations – some of which are competencies highlighted in the introduction as crucial for navigating emotionally charged situations. In challenging classroom situations, teachers’ ability to regulate their emotions has been linked to constructive teacher–student interactions, ability to manage conflicts, and maintain a positive classroom climate (Jennings & Greenberg, 2009).
Teacher educator perspective: Simulation supporting identity reflection
A simulation in which teacher students actively engage in a challenging situation provides a strong starting point for reflective discussion. Reflection on professional identity begins early in teacher education and continues throughout the programme. This simulation offered teacher students an opportunity to reflect on their pedagogical principles and on who they are as teachers when facing a challenging situation. It also brought the students together around a shared experience and prompted dialogue about how to act in difficult situations.
In fact, the simulation encouraged articulating pedagogical principles, connecting these principles to practice, and reflecting on the role of assessment. In this way, it supported preparation for both facilitating learning situations and engaging in practicum experiences. At the same time, the case offered an opportunity to explore the pedagogical potential of VR and AI-driven agents as part of teacher education.
From a teacher educator’s perspective, the case provided a novel environment for preparing teacher students to face challenging interpersonal situations. As a first iteration, the simulation can be further developed, particularly for internationally oriented teacher education contexts. The simulation was also tested by a visiting group of experienced teachers from Mozambique, whose encouraging feedback supported the continuation of the development work. Following this, the simulation was also used by other teacher education groups.
Compared to earlier simulations relying on self-created role avatars to represent both teachers and learners, this version was notably faster to prepare and implement. In the earlier approach, students created and controlled their own avatars to act out the roles in the simulation. By eliminating the need for avatar creation and management, teacher students were able to focus more directly on the pedagogical aspects of the situation. Importantly, reflective discussions focused more strongly on pedagogical aspects, whereas avatar-based simulations tended to prompt more discussion about technological possibilities. This difference may relate to the nature of the scenarios, the ease of use of technology, or the facilitation of the reflection process.
Conclusions and future
While teacher emotion regulation has been identified as central for effective classroom interaction (Aldrup et al., 2024), opportunities to practice these skills in authentic yet low-risk settings remain limited. In the future, simulation-based learning that incorporates AI agents may offer one pedagogical option to respond to this challenge. Such environments have the potential to enrich learning by providing repeatable opportunities for experimentation and reflection. Simulations can function as a “safe failure space” for rehearsing challenging classroom interactions, and research suggests that combining simulation practice with targeted coaching may better support novices’ skill development than self-reflection alone (Cohen et al., 2020).
This case illustrates one option on how VR- and AI-supported simulations can be used to create learning situations in which teacher students practice navigating emotionally challenging interactions and reflect on professional judgement in a controlled environment. This implementation appeared to support reflective engagement questions of teacher identity, and the pedagogical principles guiding decision-making in difficult situations. In this sense, the simulation potentially helped teacher students’ reflection on their teacher identities that was already underway.
The present pilot does not seek to demonstrate the effectiveness of VR- and AI-supported simulations as a comprehensive solution. Rather, it offers an exploratory example of how emotionally challenging interaction situations can be rehearsed in a controlled environment, with reflection integrated into the pedagogical design. In doing so, it responds to calls for more authentic, practice-oriented approaches highlighted for instance in a recent review by Hong et al. (2025).
Although VR simulations are increasingly used in teacher education, recent reviews suggest that research has mainly examined short-term experiences and participants’ perceptions of learning. There is still limited knowledge about how VR supports the development of emotionally demanding teaching skills, how simulations are meaningfully embedded in teacher education, or how they serve vocational teacher education in particular (Huang et al., 2023; Hong et al., 2025). As AI continues to develop, its role in teacher education may extend beyond simulating learners to supporting teacher students’ metacognition and learning processes. AI may offer low-threshold opportunities to rehearse emotionally demanding situations and to support reflection.
In this context, AI-supported simulations are understood as one complementary element within a broader teacher education and practicum process, rather than as a standalone pathway to teaching practice. Future research should examine how facilitation before, during, and after emotionally demanding simulations supports reflective learning and how such learning transfers to everyday teaching practice, particularly in vocational education contexts. In addition, preliminary observations by one of the teacher educators suggest that the perceived agency of virtual characters (human-controlled avatars vs. autonomous AI agents) may influence the focus of reflection. While this was not systematically investigated in the present study, it may offer a tentative direction for further exploration. Continued collaboration between educators, developers, and researchers is needed to co-develop simulation-based learning environments in which emotional realism, pedagogical intent and technological feasibility are meaningfully aligned.