Developing Vocational Education and Training with Artificial Intelligence

1.  Overview and key insights

Abstract

Rapid labour market and technological change – especially the rise of artificial intelligence (AI), digitalisation and the green transition – demand more agile and forward‑looking vocational education and training (VET) systems. While traditional VET curriculum and qualification development processes are rigorous and collaborative, they are also lengthy, resource‑intensive and often struggle to keep pace with evolving labour market needs.

The chapter highlights AI’s potential to transform these processes and presents the importance of this OECD study, which provides cross‑country evidence and insight and points to strategic and practical directions on how AI can support more responsive, efficient and high‑quality VET development, while maintaining the core values and collaborative foundations of VET systems.

In Brief

Overview on AI and the labour market relevance of VET

• Potential and challenges: Vocational education training (VET) curricula and qualifications are central to aligning skills supply with labour market demand, yet their development is complex, resource‑intensive and often lengthy, making it difficult to keep pace with rapidly evolving skill needs driven by digitalisation, green transitions and technological change. Poorly designed or outdated curricula contribute to qualification mismatches and skill shortages, despite the continued labour market value of qualifications. In this context, artificial intelligence (AI) offers significant potential to strengthen labour market relevance by making VET development more agile, efficient and evidence‑based. By analysing labour market data at scale, detecting emerging skill trends, supporting competency mapping and assisting with drafting and revision, AI can complement existing consultation processes, broaden the evidence base and accelerate updates – particularly in fast‑changing sectors. However, realising this potential requires careful governance, high‑quality data and human-centred approaches to avoid potential risks.

• Why this study is important: Although interest in using AI for VET development is growing, its application remains limited and fragmented. At the same time, there are currently no systematic international studies or sufficiently collected and analysed cases that identify the main challenges of introducing AI in VET development. A cross‑country perspective is therefore essential to build a robust evidence base on how AI is currently used, what works, what risks arise and under which conditions AI can contribute to the improvement of VET relevance, quality and responsiveness. By systematically documenting country practices, stakeholder experiences and emerging use cases through surveys and stakeholder interviews, this study supports peer learning and provides strategic direction on effective and secure integration of AI into VET development.

1.1. AI and the labour market relevance of VET development: Potential and challenges

Vocational education training (VET) plays a central role in connecting labour market demand with the supply of skills and in equipping learners with employability (OECD, 2023[1]). Curricula and qualifications – which refer, in this report, broadly to occupational/professional/VET standards (or regulations), qualifications and curricula – form the foundation of VET in achieving this goal. Getting their development right is thus critical: it determines how those skills required by the labour market are identified and structured, how these skills are taught, developed, assessed and certified, and how VET systems remain responsive to evolving labour market needs. When VET is designed to focus on job-specific skills and rapid school‑to‑work transitions without sufficiently integrating transversal skills and general education foundations, it may support strong initial employment outcomes but risk weakening adaptability and resilience to technological and structural change over longer careers (Hanushek et al., 2016[2]; Malamud and Pop-Eleches, 2010[3]). VET outcomes also vary across learners depending on programme design features – such as the integration of work-based learning, the use of practical training components and the structuring of skills to acquire (Attanasio, Kugler and Meghir, 2011[4]).

The process of VET curriculum and qualification development (thereafter, VET development) is necessarily rigorous and collaborative, involving dialogue between education and labour market actors and drawing on diverse stakeholder perspectives. However, it is also complex, demanding and lengthy, and varies across jurisdictions, institutions and sectors (OECD, 2025[5]). For VET policymakers, maintaining oversight while ensuring transparency, efficiency and responsiveness remains an ongoing challenge. For other VET stakeholders including social partners and providers, the process is resource‑intensive and often an additional task to their day job (OECD, 2025[5]).

The growing deployment of artificial intelligence (AI) – a machine‑based system that can, for a given set of human-defined objectives, make analysis, predictions, recommendations or decisions, with varying levels of autonomy (OECD, 2021[6]; OECD, 2024[7]) – offers a promising opportunity to address these challenges, but one that requires careful consideration. AI could help make the VET development processes more agile, efficient and evidence‑based (OECD, 2025[5]), yet to date there has been limited research, assessment or guidance on how AI can support VET development, whether within countries or across borders. This study seeks to fill that gap by exploring how AI can contribute to more timely, relevant and efficient development of VET curricula and qualifications, and at the same time identify common challenges and provide collective guidance.

The need for relevant and responsive VET curricula and qualifications: Design matters

Skills requirements are constantly evolving, driven by technological innovation, demographic shifts and global transitions such as digitalisation and the move towards greener economies (OECD, 2025[8]; OECD, 2022[9]). This dynamism places a premium on VET systems that can detect early signals of emerging skill needs and translate them into timely updates of curricula and qualifications.

From a public policy perspective, maintaining up-to-date occupational and training standards is essential to reducing skill mismatches and supporting economic growth. When VET stakeholders can rely on accurate and current standards, they are better able to design effective VET curricula and programmes that promote a more efficient allocation of human capital and help prevent unemployment (OECD, 2024[10]).

While skills mismatches often reflect gaps in individuals’ competence, qualification mismatches may stem from poorly designed or outdated curricula and qualifications – which makes the question of what to design and how to design it – particularly in terms of content relevance, learning outcomes and responsiveness to labour market change – central to VET policy. Evidence from the OECD Survey of Adult Skills (PIAAC) shows that nearly one‑third of adults experience qualification mismatch (OECD, 2024[11]). Employers frequently report shortages in job-specific skills, alongside growing demand for digital competencies (Marcolin and Quintini, 2023[12]). At the same time, education and qualification premia remain higher than skills premia, underscoring the continued importance of qualifications and the need for the associated curricula to better align with actual skill requirements (OECD, 2024[10]).

Developing and updating VET is complex, demanding and often lengthy process

While approaches to VET development vary across countries, they universally require collaboration, consultation and co‑ordination among key stakeholders – employers, employees, VET teachers, trainers and policymakers – to ensure alignment with labour market needs and educational objectives. These processes often involve reconciling diverse and sometimes competing interests. Decisions on which skills to include or exclude in VET curricula and qualifications may benefit some employers while disadvantaging others, and introducing new qualifications can reduce the appeal or value of existing ones, affecting stakeholders who have already invested in older ones. Despite these tensions, a core purpose of the process is to build consensus to serve broader goals of employability, labour market relevance and education and training quality (OECD, 2025[5]).

Beyond stakeholder dynamics, VET development is further complicated by bureaucratic or legal procedures, resource constraints and the need to balance occupation-specific competencies with transferable skills (OECD, 2024[13]; OECD, 2025[5]). At the core of this challenge are VET stakeholders’ efforts and commitment to ensure VET curricula and qualifications remain responsive to labour market demands while also supporting learners in acquiring not only specific skills but also skills that support mobility across occupations and employers, as well as progression into further education and training. As emerging technologies continue to reshape skill needs, many VET systems – along with AI developers, ICT experts and VET teachers especially those teaching IT and AI – are beginning to explore how AI tools can support more efficient and evidence‑based development processes (OECD, 2025[5]). Early applications include labour market analysis, competency mapping and automated drafting, especially in fast‑evolving sectors affected by digitalisation and the green transition (OECD, 2025[14]; OECD, 2025[5]).

Across countries, developing or revising VET curricula and qualifications to meet these diverse objectives and conditions typically takes from several months to multiple years (OECD, 2024[13]). For example, in Finland, developing a full qualification can take up to three years, with individual qualification units (modules) requiring several months. In Luxembourg and French-speaking Belgium, the process also extends up to three years, while in Latvia and Estonia, complete programme development typically takes nine months to one year. In Quebec (Canada), the average timeframe is around 18 months (OECD, 2024[13]).

Many systems report significant delays due to lengthy approval processes, limited expert availability and insufficient employer involvement, and the difficulty of integrating rapidly advancing technologies into qualification content (OECD, 2024[13]; OECD, 2025[5]). Emerging sectors, such as green hydrogen (UOK, 2023[15]), face additional pressure to deliver timely updates and, in some cases, entirely new qualifications. Ensuring that VET systems remain resilient and responsive will require not only strong and efficient stakeholder collaboration, but also the integration of innovative approaches capable of addressing this accelerating change.

AI has potential to improve VET development, and may possibly change how it works

The rapid rise of AI – especially the spread of generative AI since 2023 – has accelerated global interest in its impact on work, learning, productivity and broader societal challenges (Milanez, 2023[16]; OECD, 2022[17]; TechCouncil of Australia & Microsoft, 2023[18]). In response, countries are rapidly expanding their policy foundations for AI adoption as well as targeted use across education and training sector. Nearly all OECD countries now have national AI strategies, though their scope and ethical safeguards vary widely (OECD, 2023[19]). Several governments – for example, Australia (National AI Plan) (FSO & JSC, 2025[20]), Czechia (Digital Education Strategy, National AI Strategy), Estonia (AI Leap), Ireland (National AI Strategy; ICT Skills Action Plan; Commitment to establishing guidelines on the use of AI in education), England (National AI Strategy and AI Opportunities Action Plan) and the Netherlands (Strategic Action Plan for AI) – have launched targeted initiatives, ranging from AI‑enabled curriculum pilots to national AI-skills agendas and AI frameworks (OECD.AI, 2025[21]). These efforts reflect a growing recognition that AI can help VET systems become more responsive, efficient, inclusive and future‑ready, supporting diverse VET stakeholders, including policymakers, industry partners, teachers, trainers and learners, notably neurodivergent learners (OECD, 2026[22]).

More concretely, these developments create new opportunities to transform how VET curricula and qualifications are designed and updated, improving labour market relevance and process efficiency (Verhagen, 2021[23]). AI can analyse vacancy data, detect emerging skill trends and support both national standard‑setting and local curriculum adaptation, addressing long‑standing challenges in consultation processes by enabling more comprehensive, systemic integration of labour market information and ensuring that smaller employers and emerging sectors are not overlooked. By bringing speed, relevance and breadth of coverage, AI can help VET systems rapidly incorporate essential labour market signals and avoid a too narrow focus on inputs from a limited number of stakeholders (e.g. national VET councils with space for only a few social partners or large employers from well-established industries with resources to influence VET development) (OECD, 2025[5]).

AI can also be used to better understand the content of initial and continuing VET programmes, the skills profiles and education training pathways of young people and adults, including through the analysis of VET graduate tracking data, as well as patterns emerging from curriculum implementation. When synthesised and presented accessibly, these data sources can provide VET stakeholders with a more holistic and robust evidence base for VET development. Countries with well-established consultation processes can use AI-synthesised insights as complementary inputs that can widen and deepen understanding of skill needs and accelerate their translation into occupational standards, qualifications and VET curricula. At the same time, VET teachers and trainers and providers can be supported by AI assistants or agents that help prepare curriculum-based lesson planning, manage workloads, monitor student progress and performance, propose instructional strategies, and assist in designing or localising curriculum content based on real-time classroom insights or identified gaps in curriculum implementation and recommend differentiated resources aligned with learning goals (OECD, 2025[24]; OECD, 2026[22]). Consequently, AI use in VET development can also help improve the effectiveness of VET delivery, assessment and outcomes (see Table 1.2).

However, the potential of AI should be reviewed carefully against potential risks

While AI offers important opportunities to enhance the efficiency, flexibility and responsiveness of VET curriculum and qualification development, its use also introduces risks that require careful consideration. VET stakeholders express concerns about the uncertainties inherent in rapidly evolving AI systems, particularly given the limited evidence base on their effectiveness in VET development processes (OECD, 2025[5]). As AI models often operate as opaque systems, their outputs may be difficult to interpret or validate, underscoring the continued need for human-centre approaches (OECD-Education International, 2023[25]). In this regard, core human skills – such as critical thinking, judgment and creativity – remain essential safeguards against misuse or misinterpretation of AI‑generated insights.

A key concern is over‑reliance on or premature use of AI-generated outputs. If users become accustomed to accepting AI outputs without scrutiny, they may inadvertently embed biases or inaccuracies into VET standards, curricula or assessment frameworks. Evidence shows that AI tools trained on imperfect or unrepresentative data can reproduce and amplify those biases, influencing quality, instructional recommendations if left unchecked (OECD, 2026[26]). Such risks point to the importance of ensuring that AI supports, rather than substitutes for, expert judgement and stakeholder consensus‑building. At the same time, the access to effective AI systems and quality data to inform AI-assisted VET development is uneven across stakeholders, institutions and countries.

Effective and secure integration of AI into VET development will therefore require robust governance frameworks, high‑quality and well‑curated data, strong ethical standards and transparent decision‑making processes. It also calls for collaboration and consultation across all VET stakeholders to ensure that AI tools align with educational objectives rather than commercial incentives alone. These concerns are increasingly reflected in emerging governance and regulatory responses, such as the European Union’s Artificial Intelligence Act, which adopts a risk‑based approach and emphasises transparency, human oversight and accountability for high‑risk AI uses, including those relevant to education and training (European Parliament and Council of the European Union, 2024[27]). Although AI in teaching, learning and assessment is advancing quickly (OECD, 2026[26]), its use in VET development remains at an early stage. Continued experimentation – guided by evidence and good practices, peer learning and clear safeguards – will be crucial for realising AI’s potential while maintaining the integrity, inclusiveness and trust of VET systems (see Chapter 5, Section 5.2).

1.2. Why this cross-country study is important

Leveraging AI for the development of VET curricula and qualifications (thereafter, VET development) presents an important opportunity and a forward-thinking approach for VET systems to strengthen alignment between skills provision and labour market needs. As labour markets continue to be reshaped by demographic change, digitalisation, green transitions and accelerated technological innovation, VET systems must be forward-looking in responding to these shifts more rapidly and systematically than ever before. Harnessing AI to inform VET development can support more agile, coherent and forward‑looking VET systems, capable of providing learners with the skills needed to succeed in evolving labour markets. At the same time, barriers, risks and trade‑offs of using AI in VET development need to be assessed, anticipated and addressed. However, there are currently no systematic international studies or sufficiently collected and analysed cases that identify the main challenges of introducing AI in VET development.

This report addresses opportunities and challenges facing VET systems in developing VET curricula and qualifications

Despite growing interest in AI’s potential, the use of AI in VET development remains limited. Current applications tend to focus on labour market matching and analysis (OECD, 2023[1]; Broecke, 2023[28]), while the processes of setting national occupational standards, designing qualifications, and supporting curriculum development still largely rely on traditional approaches. This represents a crucial opportunity to harness the capabilities of AI, not just in the context of the current labour market, but as a foundational element in shaping future‑ready VET systems.

The present challenges facing VET systems in VET development are multiple. Consultation mechanisms, whether highly structured or more informal, face inherent limitations. National councils typically include a restricted number of social partners, which can inadvertently exclude smaller employers or emerging sectors where new skill needs may arise first. In systems with strong provider‑level autonomy, VET institutions face similar challenges, requiring them to gather labour market intelligence, interpret skill needs and translate these into curricula – tasks that are demanding in both time and expertise. These constraints create gaps between the speed at which labour market requirements evolve and the ability of VET systems to update or design qualifications that reflect these developments.

AI provides an opportunity to address these long-standing challenges by offering faster and broader analysis of labour market information, highlighting emerging skills, and helping VET systems to synthesise information from multiple sources. However, without evidence, guidance and shared learning, countries risk fragmented experimentation or inconsistent adoption that may fail to deliver meaningful improvements. A systematic, cross-country understanding is therefore essential.

While AI has been increasingly used across education and training for content creation, learner support and assessment (OECD, 2021[6]; OECD, 2026[26]), its application in developing and updating VET curricula and qualifications remains at an early stage. Unlike the AI use in teaching, training and learning as well as assessments, applying AI to curriculum and qualification development may raise fewer concerns about personal data, as these processes do not usually rely on learner‑level or class-level information. However, more forward-looking approaches may increasingly draw on supply-side skills data – such as information on the skills possessed by VET learners or graduates – to inform VET development, which introduces additional considerations. Questions therefore remain about transparent and secure use, governance, transparency and accountability as well as validity and quality of data. Evidence remains limited, and different perspectives – from AI developers to VET practitioners – are necessary to understand how AI can enhance agility, relevance and employability. AI’s growing capability to detect patterns, anticipate change and support decision‑making has implications for how VET systems organise expertise, validation and quality assurance (Calvino, Reijerink and Samek, 2025[29]).

This report relies largely on inputs from VET stakeholders across countries

The OECD study “Leveraging AI to Improve the Development of VET Curricula and Qualifications” aims to strengthen the evidence base on how AI is currently used – or could be used – in VET development process. It is grounded in past and ongoing global efforts to bring agility, precision, speed and efficiency to VET development (Box 1.1), VET stakeholders’ inputs across countries, and existing frameworks and principles on AI. This report is part of this study, together with a policy brief that highlights key strategic directions for VET stakeholders.

This study supports peer learning by documenting country practices, identifying emerging use cases and providing strategic directions to help countries make informed decisions, steering and support. This work was supported by the OECD Group of National Experts on VET (established as a subsidiary body of the OECD Education Policy Committee [EDPC] in April 2007 to help Members and Partners develop effective, efficient and evidence‑based VET policies) and the EDPC.

The study draws on several sources: desk research, cross‑country data collection via surveys, analysis of relevant international policies and practices, and ten in‑depth country case studies through almost 80 stakeholder interviews. All ten country examples contributed to the comparative policy and data analysis, while additional countries (25 countries in total) participated in the survey component. All the contributors were invited to international workshops held online on 3 December 2025, where they discussed preliminary findings of the study and provided initial feedback, bringing together 117 participants (89 unique) across 21 countries. Together, these elements provide a robust picture of how VET systems currently approach AI adoption, the drivers of experimentation, and the challenges encountered.

Box 1.1. Examples of efforts to bring agility, precision, speed and efficiency to VET curriculum development in the European Union

2020 Council Recommendation VET

The recommendation highlights the need for “high quality and innovative VET systems that provide people with skills that help them get or create jobs in demand on the labour market” and “working towards implementing a VET policy which equips young people and adults with the knowledge, skills and competences to thrive in the evolving labour market and society”. It calls for VET to be agile in adapting to labour market changes, including “VET curricula, programme offers and qualifications are regularly updated, as relevant, building on skills intelligence”.

Osnabrück Declaration 2020

The Declaration highlights that “Labour market and skills intelligence as well as research should inform VET decision-makers and support VET providers and practitioners in developing and adapting their VET offers to changing labour market and social needs.” One of the short-term deliverables identified in the Declaration states “Develop national and regional skills intelligence systems including skills anticipation and graduate tracking; enable social partners, decision-makers, stakeholders and providers to adapt and update VET programmes, curricula and guidelines in a timely and effective manner” (European Commission, 2024[30]).

Herning Declaration (2025)

Adopted in September 2025, the Herning Declaration on attractive and inclusive VET for increased competitiveness and quality jobs (2026-30) builds on and updates the Osnabrück Declaration. It reaffirms the commitment to agile, relevant and high‑quality VET systems, while shifting the focus from post‑pandemic recovery towards long‑term competitiveness, skills shortages and technological change. The Declaration explicitly recognises the growing role of digitalisation and emerging technologies, including artificial intelligence, and stresses the need to harness their potential responsibly to improve the responsiveness, relevance and attractiveness of VET. It also reinforces the role of social dialogue, stakeholder co‑operation and evidence‑based approaches – including skills intelligence and graduate tracking – in guiding curriculum development and system‑level reforms.

Draghi report on the future of European competitiveness

The report underlines how the integration of AI into curriculum development is crucial for Europe’s future, as it aligns with the EU’s strategy to harness AI-driven innovation to close the productivity gap, equip the workforce with necessary skills, and ensure social inclusion in the face of rapid technological advancements.

Source: European Commission (2020[31]), Council Recommendation of 24 November 2020 on vocational education and training (VET) for sustainable competitiveness, social fairness and resilience, 2020/C 417/01, https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=oj:JOC_2020_417_R_0001; European Commission (2020[32]), Osnabrück Declaration 2020 on vocational education and training as an enabler of recovery and just transitions to digital and green economies. https://www.cedefop.europa.eu/files/osnabrueck_declaration_eu2020.pdf; European Commission (2024[30]), The Future of European Competitiveness, https://commission.europa.eu/topics/competitiveness/draghi-report_en#paragraph_47059.

Methodology: Overview of survey design and country case studies

The study combines evidence from three targeted surveys and ten case studies to address the following key questions:

• How commonly AI is used in developing VET curricula and qualifications; which types of VET programmes and stages of development are involved?

• What type of data and AI tools are used?

• How do the AI-driven insights complement other information/processes?

• What have been the drivers and enabling factors for using AI? What have been the challenges?

• What is hindering AI adoption for curriculum/qualification development? How can these obstacles be overcome?

• How can AI-driven insights be effectively integrated with feedback from industry to ensure the relevance and quality of VET curricular/qualifications?

• How can the right balance be found between data and qualitative inputs, as well as between labour market-relevant skills and more general skills?

• How can the effectiveness of AI-driven curriculum and qualification development be monitored and evaluated?

• How can stakeholders be supported in making effective use of the data? How should they be informed about the limitations of the data/AI-driven insights? What level of data granularity is needed and in what format? What tools could be used to allow for easy visualisation or tailoring? Which data/information should be extracted (and how frequently)? Who should collect the data or support data collection?

These questions guided both the survey and the country interviews. The results collected from the survey analyses and case studies are integrated into this report, and all ten case studies are published in a separate report (OECD, 2026[33]). Relevant country policies and practices were reviewed through desk research.

Survey design

Surveys, focussing on ISCED levels 3‑5, were distributed to three stakeholder groups: VET policymakers, VET providers and industry partners involved in the process of developing VET curricula, qualifications or occupational standards. The VET policymakers were identified via the OECD’s Group of National Experts on VET and the OECD Education Policy Committee networks, and these policymakers distributed the survey to their relevant VET providers and industry partners.

The survey was conducted between April and November 2025 through e‑mail and SurveyMonkey. It was designed as a focussed group survey, not a population-level sample and therefore the respondents and their responses are not a statistically representative sample. Only valid responses (i.e. those that include the respondent’s name, institution, country and contact information) were retained for analysis; and after data cleaning, the final sample comprised 39 responses from VET policymakers, 47 from industry partners and 205 from VET providers. In total, 25 countries participated in at least one of the three surveys (bolded countries are included in case studies): Austria, Belgium (French Community), Canada (Nova Scotia), Costa Rica, Croatia, Czech Republic (hereafter “Czechia”), Estonia, Finland, France, Germany, Greece, Ireland, Korea, Latvia, Lithuania, Mexico, the Netherlands, Portugal, Romania, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland and the United Kingdom (England; Scotland).

Definitions of AI, curriculum/qualification development in VET, and AI use in this context were included in the surveys, as below, to ensure consistency across respondents:

• AI is a machine‑based system that can, for a given set of human-defined objectives, make analysis, predictions, recommendations or decisions, with varying levels of autonomy (OECD, 2021[6]; OECD, 2024[7]).

• Curriculum/qualification development in VET refers to “the process of designing, improving, organising and planning” VET programmes/qualifications, including: analysing education and training needs and needs for changes to qualifications; defining learning objectives and outcomes, content, activities, materials and resources, duration and level of the action, type of learning (school-based or work-based), requirements, methods, approaches, or standards (Cedefop, 2015[34]; Cedefop, 2025[35]). Box 1.2 provides details.

• Use of AI in VET curriculum and qualification development refers to using AI to develop, evaluate and update VET curricula and qualifications/standards by improving data collection and analysis, automating creation or modification of existing content in curricula/qualifications, and helping co‑ordination among relevant stakeholders. This may include using AI techniques (e.g. natural language processing, large language models, big data analysis) or AI platforms/tools (e.g. AI-driven course content analyser/generator/customiser) to map skills needs, analyse/modify/generate competency frameworks or standards, assess qualification descriptions, and structure learning outcomes. For the purposes of this study, AI applications that support personalisation of VET delivery are not included.

Box 1.2. Definition and scope of VET curriculum and qualification development in this report

VET curriculum development

“Curriculum” refers to designed and planned learning activities, while ‘programme’ refers to the implementation of these activities (Cedefop, 2015[34]). Curriculum development refers to “the process of designing, improving and planning” VET programmes including the following key components (Cedefop, 2025[35]):

• Analysis of education and training needs.

• Definition of learning objectives and outcomes, topics, content of programmes and activities, materials and resources, the duration and level of the action, type of learning (school-based or work-based), entry requirements, methods and approaches (assessment of learners, validation of learning outcomes and evaluation of programmes).

• Definition of standards (competence, educational, occupational, assessment, certification and validation standards).

The VET curriculum development includes aligning learning outcomes with National Qualifications Frameworks (NQFs), balancing between theoretical components, soft skills and practical training to gain hands-on experience in workshops, simulations or workplaces. It also includes approval by national VET authorities, evaluation and feedback from stakeholders to ensure relevance and quality.

The scope of this report does not include the actual VET delivery and curriculum implementation by VET institutions, teachers and workplaces nor building and personalising learning content/materials, teaching/assessment/evaluation methods/strategies/approaches among teachers and students.

Qualification and occupational standards development

Development of occupational standards and qualifications refers to the process of designing and structuring competences and formal credentials to meet industry, national and international education and labour market needs (OECD, 2024[10]). Occupational standards typically define the required knowledge, skills and attitudes for a given role, while qualifications certify that individuals meet these standards. Although this distinction or integration is common, it is not consistently applied across countries. Key components include competency standards and qualification levels aligned with national or regional qualification frameworks that indicate the difficulty and complexity of the knowledge and skills associated with any qualification.

Table 1.1. Occupational standards, qualifications and curriculum in VET

Aspect	Occupational standards	Qualifications	VET curricula
Focus	Define occupational competencies	Translate competencies into certification	Develop structured learning programmes
Developed by	Industry bodies, employers, sector councils	Ministries of education, labour or sectoral, qualification authorities or sectors; in some cases, employers	Ministry of education; VET institutions, curriculum designers, teachers
Purpose	Guide workforce skills and training needs, labour market planning and qualification design	Ensure formal recognition and certification of skills; define how skills are tested and certified	Provide teaching and learning plans, and planning for exams, practical assessments and assignments
Structure	Competency frameworks, skill sets, job roles	Learning outcomes, assessment criteria, certification levels	Course content, high-level lesson plans, training materials

Source: Author’s elaboration based on the OECD Surveys and Interviews conducted as part of this study (OECD, 2025[5]).

Country case studies

Ten country case studies complemented the surveys by providing contextual insights into the use of AI in VET development. Countries were selected to ensure geographic diversity, variation in VET system size and governance arrangements, differing stages of AI adoption (from nascent to more advanced) and willingness to participate in the study, including distributing stakeholder surveys and nominating participants for interviews. The case studies were conducted through semi‑structured interviews with policymakers, providers, industry partners, researchers and AI/IT experts, identified by each country focal point which was nominated by the country delegates of OECD’s Group of National Experts on VET and the OECD Education Policy Committee networks. Each focal point provided a list of 7‑10 stakeholders to interview, and a total of 76 stakeholders across ten countries participated in the interviews. One‑hour online interviews explored the same themes as the surveys but allowed deeper and wider examination of national contexts, detailed use cases, enabling factors and perceived potential risks. Transcripts formed the basis for qualitative analysis, supporting the analysis of AI use cases, the interpretation of survey findings and the formulation of VET-specific principles (OECD, 2026[36]) and policy considerations (Chapter 5). The summaries in Box 1.3 below present a concise overview of ten case studies, highlighting AI use cases in VET development. A separate report presents these case studies in details, which illustrate concrete practices, institutional arrangements and lessons learned, providing practical examples and insights to inform policymakers and VET stakeholders seeking to move from ad hoc experimentation towards more strategic, system-level use of AI in VET development (OECD, 2026[33]).

Box 1.3. Overview of ten case studies focussing on AI use cases in VET development

1. Croatia

Croatia is advancing AI skills in school curricula through the BrAIn project, yet its use in VET development remains limited and experimental. The OECD Surveys conducted as part of this study (OECD, 2025[5]) shows only 13% of VET providers use AI in VET development, below the international average (21%). Policymakers use AI informally for tasks such as analysing qualifications, aligning with EU frameworks and drafting learning outcomes, as in the case of the Directorate for Labour Market and Employment – always under human oversight. A notable pilot by the Croatian Academic and Research Network (CARNET) demonstrates more structured adoption, yet VET institutions remain cautious due to concerns over data quality, transparency and over-reliance. Barriers include limited capabilities, lack of resources and low institutional support, though interest in future use exists. Stakeholders emphasise AI as a support tool, not a replacement for expert judgement, and call for formal guidelines, training and capacity building to strengthen digital and AI literacy.

2. England (United Kingdom)

England is piloting an AI platform, SkillsCompass, developed by Skills England (formerly IfATE). This AI-enabled platform uses natural language processing and generative AI to analyse employer-defined occupational standards, real-time job vacancy data and foresight trends, identifying emerging skills needs and gaps. It can support evidence‑based decisions for updating qualifications, mapping knowledge, skills and behaviours, and signalling priorities to awarding organisations and training providers. While AI adoption remains at an early stage and human validation is essential, SkillsCompass demonstrates how AI can accelerate labour market alignment and improve system coherence in VET. Other uses in England include AI-generated duty statements, qualification mapping tools and ad hoc applications by awarding bodies and further education colleges for curriculum planning. However, interviewed stakeholders expressed concerns around data quality, stakeholder buy-in and sector-wide guidance, making AI a promising but still evolving tool in England’s VET system.

3. Estonia

Estonia is gradually integrating AI into VET curriculum and qualification development through both national initiatives and local experimentation. At the systemic level, the Estonian Qualifications Authority (EQA) is piloting AI tools to draft and refine occupational standards, describe competencies and align skills with qualification frameworks, while developing a customised AI solution and the Skills Compass platform that connects skills, occupations and labour market data for more efficient, integrated updates. AI is also being tested to automate technical validation of micro-credentials and support the country’s labour market forecasting system. At the institutional level, VET teachers and working groups for curriculum development use AI to draft learning outcomes, module descriptions and teaching materials, significantly accelerating processes for creating new VET curricula. Despite strong support from stakeholders and national AI initiatives, challenges remain around data readiness, stakeholder skills and maintaining teacher ownership, with AI seen as a tool for efficiency and translation of labour market input rather than a replacement for expert judgement.

4. Germany

Germany’s use of AI in VET curriculum and qualification development is cautious and largely informal, reflecting its emphasis on consensus-based processes, legal precision and stakeholder trust in VET. While pilots by the Federal Institute for Vocational Education and Training (BIBB) explore AI for drafting and comparing regulatory documents within secure and internal systems, systemic adoption is limited by fragmented data and strict data protection laws. AI is mainly used experimentally for tasks such as labour market analysis, competency mapping and drafting support. Expert input and judgement is still needed due to the legal weight of training regulations, risks of potential premature dissemination and the still limited quality of AI-generated output. Interviewed stakeholders see potential for efficiency gains, especially in IT-related occupations, yet stress the need for clear guidelines, robust infrastructure and capacity building. Despite growing interest and isolated initiatives, Germany prioritises transparency and human oversight, aiming to balance innovation with the integrity of its trusted VET system.

5. Ireland

Ireland’s use of AI in VET curriculum and qualification development is still exploratory and largely informal. At the national level, Ireland’s state agency for further education training (SOLAS) operates digital platforms for tracking skills and qualifications, but AI is not yet embedded systematically in curriculum generation. Quality and Qualifications Ireland (QQI) is piloting AI tools such as Microsoft Copilot for tasks including refining text, comparing standards and identifying gaps, with strict human oversight and data protection. Individual educators experiment with ChatGPT to draft exam questions and teaching resources aligned with QQI descriptors. Stakeholders emphasise that AI can improve efficiency and support content creation, but must operate within robust quality assurance frameworks, ethical guidelines and closed systems to ensure trust and compliance.

6. Korea

Korea is integrating AI into VET curriculum and qualification processes, mainly as a support tool. The Human Resources Development Service of Korea (HRDK) is testing AI to streamline updates to the National Competency Standards (NCS), applying GPT-based tools to check formatting and structural compliance of competency units. However, content development remains expert-led, as AI still lacks the contextual and sector-specific understanding required. HRDK’s customised GPT tool is currently used to check compliance with the NCS manual, ensuring consistency and efficiency without compromising content quality. Korea Research Institute for Vocational Education and Training (KRIVET) is also exploring AI for data mining and forecasting skill gaps, but only as an experimentation. Private sector initiatives use AI to analyse labour market trends for modular qualification updates. At the VET school level, AI supports project-based learning and formative assessment, while higher VET institutions train VET teachers in generative AI for content creation and updating and adjusting VET curricula. Across all cases, AI adoption remains experimental, with strong emphasis on human oversight, data security and sectoral expertise.

7. Lithuania

Lithuania’s use of AI in VET curriculum and qualification development is emerging but remains ad hoc and individual-driven. VET teachers and VET providers use tools such as ChatGPT and Copilot to draft learning outcomes, structure programmes, generate teaching materials and analyse data, while sectoral associations (e.g. ICT, woodworking, construction and hospitality) experiment with AI for document structuring, translations and competence mapping, particularly for green and digital skills. Regional stakeholders report widespread informal use across VET schools for curriculum needs analysis and updating modules, also noting limited training, infrastructure gaps and caution towards adopting AI‑supported approaches in traditionally expert‑driven curriculum development processes. Employers and private companies are more advanced, using AI for HR and short training design, but systemic integration into VET development is not yet observed. Interviewed stakeholders see AI as a support tool for efficiency and idea generation, not a replacement for expert judgment, and called for clear, relevant guidelines, secure platforms and capacity building – a recent policy initiative responding to this need is underway.

8. Mexico

Mexico’s use of AI in VET curriculum and qualification development is emerging and decentralised, driven mainly by individual curriculum designers. Free AI tools are used experimentally for labour market analysis and relevance studies, drafting content and mapping competencies, particularly in technical careers and dual VET pathways. Sectoral Coordination for Academic Strengthening (COSFAC) and General Directorate of Industrial and Technological Education (DGETI) have piloted AI for foresight studies, competency alignment and curriculum validation, while some chambers and providers apply AI for translations, structuring documents and generating didactic suggestions. Despite growing interest, AI use remains limited (6% of VET institutions) due to resource constraints, lack of training, data privacy concerns and infrastructure gaps. Stakeholders emphasise that AI complements human expertise rather than replacing it, and call for national guidelines, secure platforms and capacity building to enable effective integration into VET development.

9. The Netherlands

AI use in Dutch VET curriculum and qualification development remains at an early, experimental stage. AI is viewed as a tool to improve efficiency and alignment with labour market needs, with recognised challenges around data privacy and stakeholder Organisation for the Cooperation between VET and the Labour Market (SBB) pilots AI for automating overlap checks between qualifications and optional modules, operates a secure GPT‑4 platform (“SBB confidence. Existing initiatives are mainly driven by sector bodies and large VET institutions. The Chat”) for drafting and process support, and develops CompetentNL, a national skills platform to harmonise qualification descriptions and prepare for AI-enabled services. At the institutional level, ROC Mondrian, regional VET centre, experiments with multi‑agent AI systems for curriculum planning and learning material generation, under strict human oversight and ethical review. Sector-wide efforts by VET College Association (MBO Raad), SURF (collaborative ICT organisation for Dutch education and research institutions) and Npuls (a nationwide programme funded by the National Growth Fund to make education future‑proof through digital innovation in the Netherlands) promote secure digital environments such as EduGenAI, alongside training in AI literacy and digital didactics. EduGenAI is a secure AI platform developed with Npuls and SURF that integrates tools such as ChatGPT, Mistral, Copilot and Gemini within a protected environment, ensuring privacy, data security and adherence to public values – all outputs are validated by VET teachers. Broader rollout of EduGenAI and the finalisation of sector pacts is expected in 2026, alongside expanded training in AI literacy and digital didactics.

10. Switzerland

Switzerland’s use of AI in developing VET curricula and qualifications is at an early stage, with around 20% of VET providers reporting experimentation, mainly for curriculum design and adjustment and data analysis, while policymakers show little use or awareness. Concerns include lack of guidelines, data security and privacy, and limited AI literacy, though interest exists among providers and industry partners – particularly the ICT sector, which uses customised GPT models for modular curriculum design and exam question generation. Some VET schools have launched AI labs and ethical guidelines, and mandated VET experts and consultancies see potential for AI to accelerate competence profile development, reducing costs and timelines. However, AI is viewed strictly as a supportive tool, not a substitute for expert judgement, given Switzerland’s consensus-driven, practice‑oriented VET system. There is limited formal guidance specific to AI use in VET development yet, while some schools developed their own guidelines on AI. Interviewed stakeholders emphasised the need for clear rules, training and systemic collaboration among SERI, Cantons and professional organisations to ensure effective sector-wide integration of AI.

Source: OECD (2026[33]), Ten Country Case Studies on VET Development and AI Use, www.oecd.org/content/dam/oecd/en/publications/support-materials/2026/06/developing-vocational-education-and-training-with-artificial-intelligence_e9f76b4e/Ten-country-case-studies-on-VET-development-and-AI-use.

Terminology, scope and limitations of this study: VET development process leveraging AI

Terminology related to “vocation/al, occupation/al and profession/al” curricula, qualifications and standards, varies across countries, often due to linguistic and institutional differences. While “occupation/al” typically refers to job-specific tasks and responsibilities (OECD, 2024[10]) and “profession/al” to specialised knowledge and skills, “vocation/al” refers to education and training for specific jobs, emphasising practical skills and direct workplace application (Wedekind, 2018[37]). Therefore, “profession/al” typically refers to higher education equivalent to “vocational”, which typically refers to upper-secondary and post-secondary level education and training (OECD, 2022[38]). When examining education, training and qualification systems across countries, these terms are used in a mixed and often confusing way, particularly due to the use of translated terms in English from national language.

This report does not draw strict distinctions between these terms, as AI applications used to support their development are largely similar. For clarity, the report uses the term “VET curricula and qualifications” to refer broadly to occupational/professional/VET standards (or regulations), qualifications and curricula at ISCED levels 3‑5 (i.e. upper secondary, post-secondary and short-cycle tertiary level), while acknowledging that country notes may use national terminology. The main focus of the study is on how AI can support the design, improvement and updating of VET curricula and qualifications, including tasks such as analysing skills needs, structuring learning outcomes, mapping competency requirements, and shaping learning pathways and certification requirements.

The scope of this study is VET development within the full VET cycle, and Table 1.2 highlights how it links to other stages of the VET cycle (VET delivery, assessment and outcomes) by mapping key inputs, processes and outputs across the full VET cycle. It shows how each stage of the cycle feeds into the next – for example, labour‑market insights inform curricula and qualifications, which guide teaching and learning, leading to skills acquisition, while labour-market outcomes feed back into future VET development.

As the use of AI – particularly in VET development – remains at an early stage in most countries, this study draws not only on concrete institutional-level use cases, barriers and potential risks but also on plausible assumptions and forward-looking perspectives derived from individual experiences. Therefore, the findings should be understood as a snapshot of the use cases, challenges and potential risks identified in 2025, reflecting an evolving landscape in which AI adoption in VET development is still emerging. They do not represent definitive or final conclusions, but rather an interim picture within an ongoing process of technological and institutional change.

Table 1.2. The scope of this study on VET development and its relevance within the full VET cycle

Simplified cycle of VET (row) and tasks in each cycle (column)

	VET development	VET delivery	VET assessment	VET outcomes
Input	Skills assessment and anticipation Labour market data collection and analysis	Develop individual classroom-level or company-level VET plans based on national, local, sectoral or school-level curriculum	Developing exam questions and tasks to test skills acquired based on qualifications and curricula	VET graduates and their skills and qualifications
Process	Discussion, co‑ordination and consultation among stakeholders	Teaching/training and learning	Assess learners based on learning outcomes and assessment methods defined in qualifications and curricula	Recruitment, matching skills with labour market demand (or transition to further education)
Output	Create/modify qualification and curriculum content	Skills acquired Curriculum delivered	Validating the exam results. Certifying acquired skills.	Meeting labour market needs (the degree of which feed back into VET development as input)

Structure of the report

This report begins by setting the context and outlining why labour market relevance and responsiveness in VET development matter, highlighting AI’s potential to improve this area and its limitations – and explaining the importance, overall narrative, methodology and terminology of this study (Chapter 1). It then describes how VET curricula and qualifications are typically developed across countries and identifies key challenges that AI could help address (Chapter 2). Building on this foundation, the report analyses current and emerging AI use cases (Chapter 3), associated barriers and risks (Chapter 4), and policy considerations for effective and secure AI integration to support VET development (Chapter 5).

References

[4] Attanasio, O., A. Kugler and C. Meghir (2011), “Subsidizing Vocational Training for Disadvantaged Youth in Colombia: Evidence from a Randomized Trial”, American Economic Journal: Applied Economics, Vol. 3/3, pp. 188-220, https://doi.org/10.1257/app.3.3.188.

[28] Broecke, S. (2023), “Artificial intelligence and labour market matching”, OECD Social, Employment and Migration Working Papers, No. 284, OECD Publishing, Paris, https://doi.org/10.1787/2b440821-en.

[29] Calvino, F., J. Reijerink and L. Samek (2025), “The effects of generative AI on productivity, innovation and entrepreneurship”, OECD Artificial Intelligence Papers, No. 39, OECD Publishing, Paris, https://doi.org/10.1787/b21df222-en.

[35] Cedefop (2025), Terminology of European education and training policy: Curriculum development, https://www.cedefop.europa.eu/en/tools/vet-glossary/glossary/curriculumontwikkeling.

[34] Cedefop (2015), The role of modularisation and unitisation in vocational education and training, Luxembourg: Publications Office. Cedefop working paper; No 26., https://doi.org/10.2801/38475.

[30] European Commission (2024), The Future of European Competitiveness, https://commission.europa.eu/topics/competitiveness/draghi-report_en#paragraph_47059.

[31] European Commission (2020), Council Recommendation of 24 November 2020 on vocational education and training (VET) for sustainable competitiveness, social fairness and resilience, 2020/C 417/01, https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=oj:JOC_2020_417_R_0001.

[32] European Commission (2020), Osnabrück Declaration 2020 on vocational education and training as an enabler of recovery and just transitions to digital and green economies, https://www.cedefop.europa.eu/files/osnabrueck_declaration_eu2020.pdf.

[27] European Parliament and Council of the European Union (2024), Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act).

[20] FSO & JSC (2025), FSO and Microsoft join forces to accelerate AI upskilling across the vocational education sector, Future Skills Organisation, https://www.futureskillsorganisation.com.au/wp-content/uploads/2025/08/FSO-Skills-Accelerator-AI-Press-Release-13082025.pdf.

[2] Hanushek, E. et al. (2016), “General Education, Vocational Education, and Labor-Market Outcomes over the Lifecycle”, Journal of Human Resources, Vol. 52/1, pp. 48-87, https://doi.org/10.3368/jhr.52.1.0415-7074r.

[3] Malamud, O. and C. Pop-Eleches (2010), “General Education versus Vocational Training: Evidence from an Economy in Transition”, Review of Economics and Statistics, Vol. 92/1, pp. 43-60, https://doi.org/10.1162/rest.2009.11339.

[12] Marcolin, L. and G. Quintini (2023), “Measuring skill gaps in firms: the PIAAC Employer Module”, OECD Social, Employment and Migration Working Papers, No. 292, OECD Publishing, Paris, https://doi.org/10.1787/903c19c9-en.

[16] Milanez, A. (2023), “The impact of AI on the workplace: Evidence from OECD case studies of AI implementation”, OECD Social, Employment and Migration Working Papers, No. 289, OECD Publishing, Paris, https://doi.org/10.1787/2247ce58-en.

[22] OECD (2026), AI to Support Neurodivergent Learners in Vocational Education and Training, OECD Publishing, Paris, https://doi.org/10.1787/718d7522-en.

[36] OECD (2026), “Five principles for the effective use of AI in vocational education and training development”, OECD Publishing, Paris, https://doi.org/10.1787/ee8c8e8e-en.

[26] OECD (2026), OECD Digital Education Outlook 2026: Exploring Effective Uses of Generative AI in Education, OECD Publishing, Paris, https://doi.org/10.1787/062a7394-en.

[33] OECD (2026), Ten Country Case Studies on VET Development and AI Use, OECD, Paris, https://www.oecd.org/content/dam/oecd/en/publications/support-materials/2026/06/developing-vocational-education-and-training-with-artificial-intelligence_e9f76b4e/Ten-country-case-studies-on-VET-development-and-AI-use.

[5] OECD (2025), Developing vocational qualifications and curricula with AI: Surveys and stakeholder interviews, Unpublished.

[8] OECD (2025), Empowering the Workforce in the Context of a Skills-First Approach, OECD Skills Studies, OECD Publishing, Paris, https://doi.org/10.1787/345b6528-en.

[24] OECD (2025), “OECD Teaching Compass: Reimagining teachers as agents of curriculum changes”, OECD Education Policy Perspectives, No. 123, OECD Publishing, Paris, https://doi.org/10.1787/8297a24a-en.

[14] OECD (2025), Vocational Education and Training and the Green Transition in Finland, OECD Reviews of Vocational Education and Training, OECD Publishing, Paris, https://doi.org/10.1787/4d29a34a-en.

[10] OECD (2024), Agile Occupational and Training Standards for Responsive Skills Policies, Getting Skills Right, OECD Publishing, Paris, https://doi.org/10.1787/bacb5e4a-en.

[11] OECD (2024), Do Adults Have the Skills They Need to Thrive in a Changing World?: Survey of Adult Skills 2023, OECD Skills Studies, OECD Publishing, Paris, https://doi.org/10.1787/b263dc5d-en.

[7] OECD (2024), “Explanatory memorandum on the updated OECD definition of an AI system”, OECD Artificial Intelligence Papers, No. 8, OECD Publishing, Paris, https://doi.org/10.1787/623da898-en.

[13] OECD (2024), VET and the green transition background questionnaire.

[1] OECD (2023), Building Future-Ready Vocational Education and Training Systems, OECD Reviews of Vocational Education and Training, OECD Publishing, Paris, https://doi.org/10.1787/28551a79-en.

[19] OECD (2023), Government at a Glance 2023, OECD Publishing, Paris, https://doi.org/10.1787/3d5c5d31-en.

[17] OECD (2022), “Harnessing the power of AI and emerging technologies: Background paper for the CDEP Ministerial meeting”, OECD Digital Economy Papers, No. 340, OECD Publishing, Paris, https://doi.org/10.1787/f94df8ec-en.

[38] OECD (2022), Pathways to Professions: Understanding Higher Vocational and Professional Tertiary Education Systems, OECD Reviews of Vocational Education and Training, OECD Publishing, Paris, https://doi.org/10.1787/a81152f4-en.

[9] OECD (2022), Skills for the Digital Transition: Assessing Recent Trends Using Big Data, OECD Publishing, Paris, https://doi.org/10.1787/38c36777-en.

[6] OECD (2021), OECD Digital Education Outlook 2021: Pushing the Frontiers with Artificial Intelligence, Blockchain and Robots, OECD Publishing, Paris, https://doi.org/10.1787/589b283f-en.

[21] OECD.AI (2025), OECD.AI Policy Navigator, https://oecd.ai/dashboards (accessed on 10 April 2026).

[25] OECD-Education International (2023), Opportunities, Guidelines and Guardrails on Effective and Equitable Use of AI in Education, https://www.oecd.org/content/dam/oecd/en/about/projects/edu/smart-data-and-digital-technology-in-education/Opportunities,%20guidelines%20and%20guardrails%20for%20effective%20and%20equitable%20use%20of%20AI%20in%20education.pdf.

[18] TechCouncil of Australia & Microsoft (2023), Australia’s Generative AI opportunity, https://techcouncil.com.au/wp-content/uploads/2023/07/230714-Australias-Gen-AI-Opportunity-Final-report-vF4.pdf.

[15] UOK (2023), Development of professional qualifications: Final report of the working group, http://urn.fi/URN:ISBN:978-952-263-738-3.

[23] Verhagen, A. (2021), “Opportunities and drawbacks of using artificial intelligence for training”, OECD Social, Employment and Migration Working Papers, No. 266, OECD Publishing, Paris, https://doi.org/10.1787/22729bd6-en.

[37] Wedekind, V. (2018), Vocational versus occupational qualifications: Is there a difference, and does it make a difference?, The South African Qualifications Authority.