It is well accepted that artificial intelligence (AI) will have a disruptive impact on working life and skill requirements, but experts differ in their estimations of the magnitude and the dynamics of these transformations (Humlum & Vestergaard, 2025). The release of ChatGPT provided AI accessibility to the general public and has impacted the educational sector to such an extent that scholars are asking for a paradigm shift in curriculum design, instructional practices and evaluation methods (articles in UNESCO, 2025). A widespread concern in the growing research field of AI in education (AIED) is the gap between the proliferation of studies on technological advancements and the lack of critical research on their implementation in pedagogical practice (Ifenthaler et al., 2024). This state of the art is particularly true for vocational education and training (VET), which on the one hand, is more exposed to AI-infused transformations in working life (Wuttke et al., 2020) while on the other hand, represents an under-researched field in AIED (Bekiaridis & Attwell, 2024; Petridou & Lao, 2024). To counter this knowledge gap, more research is needed to develop a timely AI vocational pedagogy that is grounded in VET teachers’ AI-mediated practices and professional competences and that provides well-founded guidelines (a didactic) for vocational training quality and teachers’ competence development in an AI era. Reviews of policy-oriented and research models of teachers’ professional AI competences (Mikeladze et al., 2024) have concluded that the abstractness of such frameworks may seriously reduce their relevance in practice. Thus, Filo and Mor (2024) propose a turnaround strategy that complements the theory-driven and normative top-down approaches with bottom-up contextual studies of teaching practices in AI contexts.  The present exploratory study follows this suggested redirection that seems especially valid given the complexity of VET.

We study how in-service VET teachers in the Norwegian education programme Information Technology and Media Production (ITMP) in upper secondary school integrate AI in the two-year school-based part of VET. The students (age 15–18) learn about user support, IT networks, IT security and programming, storytelling, photography, film, and visual communication, and in the school-based part, they engage in shorter projects with relevant companies. The school-based education prepares the students for two years of apprenticeship in training establishments within five recognised trades.

Our main research questions were: How does the use of AI in the ITMP programme impact the subject-specific and general didactical practices of vocational teachers? How are the professional digital competencies of vocational teachers transformed in the context of AI-mediated VET?

The objective is to examine changes in teaching practices, teacher roles and student progression associated with AI use and to document teachers’ pathways for competence development and identify institutional barriers and enablers. An ultimate objective is to provide empirically grounded recommendations for developing VET-specific AI didactics and policy.

The theoretical framework for this study is grounded in a socio-material perspective on learning and training, focusing on the co-constitution of individual cognition, social interaction, and performativity with artefacts understood as tools designed for human purposes (Säljö, 2019). As AI reshapes teaching by automating assessment and feedback, supporting personalized learning, and reconfiguring teachers’ roles as facilitators, orchestrators, and co-learners, these transformations are interacting with teachers’ professional identities and institutional norms. The framework of boundary-crossing and connectivity (Kyndt et al., 2022) provides a dynamic understanding of the transfer of knowledge and skills across the divide between work and school in VET that foregrounds the entanglement of technology, knowledge, and practice. This lens helps explain how teacher roles and didactic practices co-evolve with AI affordances rather than change in isolation.

In order to capture rich, practice-driven insights from AI-active practitioners, this exploratory study purposively oversampled pioneering teachers in the ITMP programmes of two schools. These served as sub-cases within a broader networked-case setting (schools and industry partners in the ITMP-field). In line with designs of multiple case studies (Stake, 2013), we collected data using multiple methods: (1) semi-structured personal interviews with teachers in the two schools (N= 4), transcribed, (2) researchers’ field notes from observations in the schools and two network meetings, and (3) the collection of documentary data (lesson plans, presentations from teachers and companies in the two network meetings. Interview and observation guides were informed by prior research on AI-integrated teaching and VET teachers’ professional digital competences, and organised around three analytic categories: (1) subject-specific AI use, (2) teachers’ professional roles and pedagogy in an AI context, and (3) professional development of pioneering teachers. Data analysis followed an abductive thematic approach (Thompson, 2022). Researchers first identified patterns across datasets, performed tentative coding aligned with the three categories, and grouped codes into candidate themes by seeking cross-cutting concepts and explanatory links (e.g., subject-specific digital competencies as drivers of innovative practice). Themes were iteratively reviewed across all data sources to triangulate consistent and divergent findings. Generative AI (ChatGPT-like) tools were used as a complementary aid in qualitative analysis (Yang & Ma, 2025): researchers iteratively prompted the model to summarise initial interview themes and to surface possible relationships. To mitigate hallucination and bias risks, outputs were cross-checked against manual analytic notes; AI-derived suggestions were treated as prompts for further scrutiny rather than as standalone interpretations. This assisted transparency in coding criteria and provided alternative patterns to test against researchers’ domain knowledge. Analytic distinctions were explicitly maintained between observed AI practices and participants’ stated views on AI and professional digital competencies. The study is intended to be a design base for planned replications in other vocational fields (health, construction/technology), emphasizing generalisability through comparative future work.

The study addresses three interrelated themes: vocational didactics integrating AI, general pedagogical transformations in AI-practices and teacher competence development in an AI-context. Vocational didactic integration. AI is experienced as embedded in occupational tools and practices (e.g. code-generation aids). It supports the development of new vocational competences by reducing routine manual tasks, enabling higher-order creative work, and linking theory to practice (e.g., “Startup with AI” projects). Successful use depends on subject-specific knowledge, precise prompting skills, and critical interpretation of AI outputs. Ethical concerns (privacy, licenses) are addressed pragmatically but reveal equity issues when students rely on privately funded AI licenses. General pedagogical effects. Our analyses show, in line with previous studies (Olivier and Weilbach, 2024; Prieto et al., 2025) that AI functions as an “assistant teacher” across planning, in-class support, and formative assessment—facilitating immediate feedback, differentiated pathways, and to some extent enhancing student self-regulation. Assessment practices shift toward process documentation and reflection to guard against superficial AI use. AI also enhances inclusion (e.g., supports learners with dyslexia) while posing risks of widening achievement gaps. Teacher competence development. Competence growth is largely self-directed—driven by motivation, informal networks, industry contacts, and experimentation. Teachers increasingly act as co-learners with students, yet institutional support and formalised professional development were limited and may reflect an inertia in the face of rapid technological changes in schools and work. Our study calls for policy and resource responses to ensure equitable access, integrate ethics and critical thinking in AI pedagogy, and to formalise AI competence frameworks across vocational fields. Comparative and longitudinal studies are recommended to validate and extend these provisional patterns of vocational teachers in the field of ITMP as AI entrepreneurs. We are presently outlining a proposal with such extensions including international research collaboration with other European countries.

Bekiaridis, G. & Attwell, G. 2024. Integrating Artificial Intelligence in Vocational and Adult Education: A Supplement to the DigCompEdu Framework. UbiquityProceedings, 4(1), 20. https://doi.org/10.5334/uproc.142 Filo, Y. & Mor, Y. (2024) An Artificial Intelligence Competency Framework for Teachers and Students: Co-created With Teachers. European Journal of Open, Distance and E-Learning, 26. https://doi.org/10.2478/eurodl-2024-0012 Ifenthaler, D. et al. (2024) Artificial Intelligence in Education: Implications for Policymakers, Researchers, and Practitioners. Technology, Knowledge and Learning, 29, 1693–1710. https://doi.org/10.1007/s10758-024-09747-0 Humlum, A. and Vestergaard, E. (2025) Large Language Models, Small Labor Market Effects . Cambridge. National Bureau of Economic Research (NBER). https://doi.org/10.3386/w33777 Kyndt, E. Beausaert, S.& I. Zitter (Eds) (2022). Developing connectivity between education and work. Routledge. https://doi.org/10.4324/9781003091219-2 Mikeladze, T., Meijer, P. C., & Verhoeff, R. P. (2024). A comprehensive exploration of artificial intelligence competence frameworks for educators: A critical review. European Journal of Education, 59(3). https://doi.org/10.1111/ejed.12663 Olivier, C. & Weilbach, L. (2024). Enhancing Online Learning Experiences: A Systematic Review on Integrating GenAI Chatbots into the Community of Inquiry Framework. Lecture Notes in Computer Science, 14907. https://doi.org/10.1007/978-3-031-72234-9_7 Petridou; E. & Lao, L. (2024) Identifying challenges and best practices for implementing AI additional qualifications in vocational and continuing education: a mixed methods analysis, International Journal of Lifelong Education, 43(4), 385-400 https://doi.org/10.1080/02601370.2024.2351076 Prieto, A., Romero, A. & Bellas, F. (2025). Advancing Robotics Education: Integrating Large Language Models for Natural Language Programming in VET. Lecture Notes in Computer Science, 15347. https://doi-org.ezproxy.uio.no/10.1007/978-3-031-77738-7_44 Shen, Y., Liu, Q., Zhang, K. & Zou, R. (2023) The Application of Artificial Intelligence Technology in Personalized Teaching of Vocational Education. IEEE. https://doi.org/10.1109/ITME60234.2023.00172 Stake, R.E. (2013). Multiple Case Study Analysis. Guilford Press. ISBN 9781593852481 Säljö, R. (2019). Materiality, Learning, and Cognitive Practices: Artifacts as Instruments of Thinking. In: Cerratto et al (eds) Emergent Practices and Material Conditions in Learning and Teaching with Technologies. Springer, Cham. https://doi.org/10.1007/978-3-030-10764-2_2 Thompson, J. (2022). A Guide to Abductive Thematic Analysis. The Qualitative Report, 27(5), 1410-1421. https://doi.org/10.46743/2160-3715/2022.5340 UNESCO (2025) AI and the future of education. Disruptions, dilemmas and directions. UNESCO. https://doi.org/10.54675/KECK1261 Wuttke, E., Seifried, J., & Niegemann, H. (Red.). (2020).Vocational Education and Training in the Age of Digitization. Verlag Barbara Budrich. https://doi.org/10.3224/84742432 Yang, Y. & Ma, L. (2025) Artificial intelligence in qualitative analysis: a practical guide and reflections based on results from using GPT to analyze interview data in a substance use program. Qual Quant. https://doi.org/10.1007/s11135-025-02066-1