Beyond AI Skills: Generative Artificial Intelligence and Work-Based Learning in Craft and Creative VET

Context: Current debates on Vocational Education and Training (VET) increasingly emphasise the need to equip learners and workers with artificial intelligence (AI) skills, particularly in cultural and creative sectors. European policy initiatives often frame this challenge through competence development, qualification reform, and employability agendas. While important at system level, such perspectives provide limited insight into how vocational learning is enacted in everyday work settings where knowledge is socially, materially, and historically situated. Research on work-based learning has long demonstrated that vocational competence develops through participation in occupational practice, mentoring relations, and engagement with workplace affordances (Billett, 2001; Lave & Wenger, 1991). This perspective is particularly relevant in artisanal and creative occupations, where learning relies on embodied judgement, tacit knowledge, and situated problem-solving (Sennett, 2008; Marchand, 2016). Against this background, the study examines how generative AI is entering work-based learning practices in Tuscany’s craft and creative sector. Approach: This first-stage study adopts a qualitative exploratory multi-source design with focused ethnographic elements. Over a twelve-month period (April 2025 – April 2026), data were drawn from one-day focused visits to four settings — Santa Chiara Lab – FabLab Siena, FabLab Conventino – Officina Creativa Florence, Manifattura Tabacchi Florence, and the Mostra Internazionale dell'Artigianato (MIDA) — combining workplace observation, informal conversational encounters with craft practitioners, and documentary and digital materials. A prior empirical mapping of hybrid craft and digital workplaces in Tuscany (Sgorla, 2025) provided the basis for purposive site selection. Data were analysed through iterative thematic coding informed by sensitising concepts from work-based learning, situated learning, and sociomaterial theory. Findings: Early findings indicate that generative AI is being incorporated unevenly across craft settings, used more frequently for ideation, marketing and design support than for core manual production. Learning about AI occurs informally through self-directed experimentation and peer exchange rather than formal training. Tensions emerge between innovation narratives and concerns regarding authenticity, authorship, and occupational identity. The notion of «the hand» (la mano) functions as a boundary marker between tasks where AI may support the creative process and those where embodied execution remains non-negotiable. Craft practitioners evaluate AI-generated outputs through sensory, aesthetic, and practical judgement developed over years of material engagement, positioning AI as raw material to be filtered rather than as a replacement for craft knowledge. Conclusions: The study contributes to VET research by moving beyond competence-based framings of AI and examining how technological change is mediated through situated vocational practice. Conceptually, it advances the notion of hybrid learning ecologies (Sgorla et al., 2025), understood as learning environments shaped through the interaction of embodied workplace practice, mentoring and peer relations, institutional training arrangements, public sectoral networks and events, and AI-mediated knowledge resources. The Tuscan case suggests that a critical capacity for craft workers may reside not merely in the ability to use AI tools, but in the evaluative competence to contextualise and integrate AI outputs within situated practice.