Generative artificial intelligence (GenAI) has gained traction as both a tool for and a topic of learning. However, the potentialities ascribed to this technology – such as being a “partner”, “tutor” or “facilitator” – starkly contrast with the reality of educators only partially knowing how to apply it meaningfully to support students’ learning. This seems particularly relevant in vocational education and training (VET), where professional learning is often less formalized than in other branches of the educational system (Zhou et al., 2022). Moreover, overly optimistic claims about the instrumental nature of GenAI for learning gains and successes can easily lead to adoptions that supersedes pedagogical deliberation (Chan, 2025).

In a current research project, we work with teachers in a specific field of vocational education in the Nordics, namely Social and Health Care education, which combines nursing, support, and practical assistance for, for example, the elderly. A growing number of students in this field originate from non-native-speaking countries, and the majority of them are women (Aarkrog, 2020). 

In the project,researchers, teachers and didactical consultants collaborate to strengthen the participation of these students, both in school and subsequently in the Danish labor market. The project builds on the thesis that GenAI as a technology with large language-based capacities can be applied as a resource for reducing the language- and participation-related challenges that bilingual women experience in Social and Health Care education.  Over the last two years there has been increasing recognition of the deep interconnection between vocational practice and digitalization, specifically for migrant and multilingual workers (Bradley et al., 2025; Lindström & Hashemi, 2019), along with a growing awareness that pedagogies integrating GenAI as a specific tool to support these groups are both overlooked and needed (e.g., Creely & Barnes, 2025).

As one entry point to achieving better participation of bilingual students, the project engages teachers through an action-learning and design-thinking inspired workshop series (e.g. Lindvig & Mathiasen, 2020; see also Schmitt & Brutzer, 2025). Over six-months, teachers meet to develop and refine custom-made chatbots intended to support bilingual students’ learning and participation. Teachers themselves define the pedagogical challenge and how they wish to address it (e.g., developing a chatbot that supports subject-specific language training, understanding Social and Health Care concepts, or providing an AI study-buddy for questions about life and health care in the Nordics). Between meetings, teachers test the chatbots under real-life classroom conditions, resulting in solutions that vary in breadth and depth.

It has become evident through the project, which will continue to work with teachers in 2026, that not all participants succeed in developing fully functional and scalable chatbots. This seems to be due to various reasons, both external factors such as lack of time and support in the school ecosystem, but also to questions which learning and support needs are being identified for bilingual students. Nevertheless, even when they do not succeed technically, participants engage in meaningful learning about the potential and challenges of GenAI as an element in their teaching.

In this presentation we will dive deeper into the question what types of learning and learning processes participants experience in the tension between creating a deliverable product and attending to their own interests, problem-definitions and contextual constraints. Through the lens of Cultural Historical Activity  (CHAT, e.g. Engeström, 1987) we analyze the contradictions induced by both the workshop design and the affordances of GenAI.

The projects collects data through a methodology inspired by ethnographic action research (Tacchi et al., 2023), recognizing that all participants, teachers and researchers alike, act as legitimate co-producers and co-owners of knowledge (Rohwedder et al., 2024). The research teams follows the unfolding dynamics as GenAI becomes part of teachers’ pedagogical practice and consequently also of students’ learning – thus acting as a “social-cultural animator” (Tacchi et al., 2023, p. 27). Researchers share their observations, insights and analytical perspectives to inspire, encourage and support teachers’ leaning and development work. This encompasses an open and inductive research approach with multi-representational documentation of processes and products, including video and audio recordings, documentation of artefacts (both chatbots and development materials), structured interviews, informal conversations and accompanying fieldnotes collected throughout the testing of solutions under real-life classroom conditions. Specifically, we conduct several rounds of Observiews - a data-gathering approach combining researchers’ classroom observations with immediate post-observation interviews and joint reflection (Kragelund, 2013). Observiews foster mutual reflection, where interviewees gain new insight into their own practice through the researcher’s questions, while researchers deepen their understanding through participants’ reflections. The theoretical angle to analyze the learning dynamics participants experience CHAT and the notion of expansive learning (Engeström & Sannino, 2010). At the heart of CHAT is the activity system—a framework for understanding human actions as socially and culturally embedded. Expansive learning occurs when elements in the activity systems are being transformed to resolve contradictions that limit the execution of the activity. We apply framework analysis and inductive coding guided by the elements of the activity system (e.g., Gale et al., 2013). Our current analysis draws on materials (video/audio recordings, transcripts, field observations, transcripts of three Observiews as well as a quantitative follow-up questionnaire) gathered during the first round of the workshops with n=26 participants in 2025. A second round with 19 participants begins in February 2026, with additional Observiews and interviews to follow.

The preliminary analysis revealed so far that teachers experience tensions in terms of positioning GenAI as an object of their own learning (‘learning about GenAI’) versus as a tool supporting bilingual students’ language- and participation-related learning (‘learning with GenAI’). In the first case, the focus lies on acquiring knowledge and competences about the workings and technicalities of GenAI; in the second teachers focus on students’ learning needs as identified in the initial problem formulation. Event teachers who perceive their chatbot-development as successful remain at least partly in doubt how well their tools support students, partly due to only limited implementation in their schools. However, knowledge and competences acquired during the workshops are being seen as distributed with other colleagues and other actors in schools, as well as supporting pedagogical work around the use of GenAI in Social and Health Care educations. These findings highlight the complex nature of combining teacher professional development on GenAI with hands-on product development aimed at supporting students. While production-based learning is not new (e.g., Poulsen, 2011), our analysis suggests that a technology as pervasive and ethically ambivalent as GenAI eventually may require professional development approaches that deliberately integrate use, reflection and critical examination of the technology’s affordances and limitations.

Aarkrog, V. (2020). The standing and status of vocational education and training in Denmark. Journal of Vocational Education & Training, 72(2), 170–188. Bradley, L., Guichon, N., & Kukulska-Hulme, A. (2025). Migrants’ and refugees’ digital literacies in life and language learning (Special Issue). ReCALL, 37(Special Issue 2), 147–156. Chan, S. (2025). Guidelines and Recommendations for the Integration of Gen AI into VET Learning. In S. Chan (Ed.), Artificial Intelligence in Vocational Education and Training: Understanding Learner and Teacher Perspectives on the Integration of Generative AI through Participatory Action Research (pp. 179–200). Springer Nature Singapore. Creely, E., & Barnes, M. (2025). Exploring attitudes to generative AI in education for English as an additional language (EAL) adult learners. ReCALL, 37(2), 174–190. Engeström, Y. (1987). Learning by Expanding: An Activity—Theoretical Approach to Developmental Research. Orienta-Konsultit. Engeström, Y., & Sannino, A. (2010). Studies of expansive learning: Foundations, findings and future challenges. Educational Research Review, 5(1), 1–24. Gale, N. K., Heath, G., Cameron, E., Rashid, S., & Redwood, S. (2013). Using the framework method for the analysis of qualitative data in multi-disciplinary health research. BMC Medical Research Methodology, 13(1), 117. Kragelund, L. (2013). The obser-view: A method of generating data and learning. Nurse Researcher, 20(5), 6–10. Lindström, N. B., & Hashemi, S. S. (2019). Mobile technology for social inclusion of migrants in the age of globalization. A case study of newly arrived healthcare professionals in Sweden. The International Journal of Technology, Knowledge, and Society, 15(2), 3–21. Lindvig, K., & Mathiasen, H. (2020). Translating the Learning Factory model to a Danish Vocational Education Setting. Procedia Manufacturing, 45, 90–95. Poulsen, M. (2011). Learning by producing. In M. Poulsen & E. Køber (Eds), The GameIT Handbook. A framework for game based learning pedagogy (pp. 87–103). http://www.projectgameit.eu/. Rohwedder, A.-B. N., Møller, B., & Kordovsky, J. (2024). Creating knowledge equity and a social learning space in practitioner-researcher collaborations: A didactic perspective. Nordic Journal of Vocational Education and Training, 14(3), 114–137. Schmitt, C., & Brutzer, A. (2025). Generative artificial intelligence in vocational education and training: A framework for sustainable teacher competence development. Tacchi, J., Slater, D., & Hearn, G. (2023). Ethnografic action research. United Nations Educational UNESCO. Zhou, N., Tigelaar, D. E. H., & Admiraal, W. (2022). Vocational teachers’ professional learning: A systematic literature review of the past decade. Teaching and Teacher Education, 119, 103856.