Developing Vocational Education and Training with Artificial Intelligence

5. Policy considerations

Abstract

There are barriers and risks associated with the use of artificial intelligence (AI) in vocational education and training (VET) curriculum and qualification development, as identified in the previous chapter (Chapter 4). Building on this analysis, this chapter concludes by outlining five guiding principles for the use of AI in VET development and providing policy considerations on how these barriers and risks are being or could be addressed, as well as ways to operationalise those five principles.

In Brief

Principles and policy considerations for effective and secure use of AI to support VET curriculum and qualification development

How should countries address barriers and risks associated with the AI in VET development?

• While the use of artificial intelligence (AI) in curriculum and qualification development in vocational education and training (VET) presents significant opportunities, there are barriers and risks associated with it, including data protection, intellectual property, security and the quality and reliability of AI‑generated outputs. Access to AI tools and high-quality, interoperable and machine‑readable data remains uneven across VET stakeholders, while fragmented governance arrangements, legal restrictions and weak data infrastructures limit scalability, comparability and system-level impact. VET stakeholders also highlight risks of over-reliance on AI, weakening professional judgement, multi-stakeholder governance and accountability when AI is not carefully embedded within established validation and decision making processes (see Chapter 4).

• Given that VET development has unique institutional characteristics to be considered when applying AI through negotiated, regulated and multi stakeholder processes, with strong signalling effects for learners and employers, existing general AI principles are insufficient. This points to a clear need for VET‑specific guidelines to ensure AI strengthens rather than undermines collective decision‑making, trust and quality. In response, this study distils five guiding principles: human‑centred use, equity and diversity, accountability, transparency, and strong data quality, security and protection.

• Because challenges associated with AI use in VET development are not only technical but also organisational, pedagogical and cultural, these principles require concrete, actionable policy measures to be operationalised. Effective, secure and inclusive AI integration depends on a clear purpose, proportionate regulation, robust data governance, transparent allocation of responsibilities, and sustained human‑centred process design and oversight. Moving beyond isolated pilots or individual-based use towards system‑level impact therefore requires coherent national and sectoral strategies, targeted capacity‑building and access support, clear guidelines, and interoperable data ecosystems that balance innovation with quality, equity, trust and regulatory stability.

5.1. The uniqueness of VET development and the need of VET-specific AI principles

As in other policy areas, the effective and secure use of artificial intelligence (AI) to support the development of curriculum and qualification in vocational education and training (VET) (thereafter, VET development) depends on a range of enabling conditions, including clear strategic direction, appropriate policy and regulatory frameworks, effective governance arrangements, investment in data and infrastructure, and sufficient stakeholder capacity and support. At the same time, VET development has unique institutional characteristics to be considered when applying AI, given that VET curricula and qualifications are developed through negotiated, regulated and multi‑stakeholder processes, often embedded in legal frameworks and quality‑assurance arrangements and involving public authorities, social partners, sector bodies and providers. Decisions taken in these processes carry strong signalling effects for learners and employers and have direct implications for labour‑market outcomes and public trust in VET. There is a clear need for VET-specific guidance on the effective, secure and inclusive use of AI in VET development processes, as existing international AI principles – while providing an important foundation – do not fully account for the institutional realities of VET systems.

In this context, this chapter presents guiding principles and policy considerations that are tailored to unique features of VET. These are grounded in practical realities of countries participating in the OECD Surveys and Interviews conducted as part of this study (OECD, 2025[1]) and are informed by many existing OECD and EU principles, frameworks and guidance (Varsik and Vosberg, 2024[2]; OECD, 2019[3]; OECD, 2026[4]; European Parliament and Council of the European Union, 2016[5]; European Parliament and Council of the European Union, 2024[6]; G7, 2024[7]; OECD-Education International, 2023[8]).

Building on the OECD Surveys and Interviews conducted as part of this study (OECD, 2025[1]) and existing OECD and EU principles, frameworks and guidance, the five principles (OECD, 2026[9]) are distilled to support VET stakeholders on the effective, secure and inclusive use of AI in VET development. They are in line with the existing AI principles, frameworks and guidance, while providing VET-specific insight. They clarify how AI should support collective expert judgement and decision making rather than replace it.

1. Human-centred use emphasises that AI should function as a support tool for VET development, assisting with analytical or technical tasks while final judgement, validation and responsibility remain firmly with human experts and stakeholders.

2. Diversity and inclusiveness focusses on ensuring that AI use does not advantage only well-resourced stakeholders, but instead supports broad and inclusive participation in VET development, including SMEs, smaller providers and stakeholders with lower digital or AI capacity.

3. Accountability clarifies that responsibility for decisions, processes and outcomes in VET development must remain clearly assigned to VET stakeholders, even when AI is used, preventing ambiguity about who owns and approves curricula and qualifications.

4. Transparency requires that stakeholders understand when and how AI is used in VET development, for what purpose, on which data and with what limitations, helping to maintain trust, supports continuous improvement, and avoid over-reliance on AI-generated outputs.

5. Data quality, security and protection highlights the need for high quality, relevant and well- governed data, alongside strong safeguards for sensitive educational and labour market information, ensuring that AI use complies with legal requirements and preserves trust in VET systems.

The policy considerations below translate these principles into strategic and operational directions for policymakers and VET stakeholders, addressing how enabling frameworks, concrete governance arrangements, capacity‑building measures and data infrastructures can be designed to embed these principles in practice. In this way, the principles provide a stable, universal values‑based compass for effective, secure and inclusive AI use in VET development, while the policy considerations outline the concrete pathways and actionable directions through which they can be implemented, balanced and adapted over time, taking into account national contexts, regulatory environments and system capacities.

5.2. Policy considerations

The following policy considerations translates the analytical findings and evidence presented in Chapters 2‑4 to guide VET stakeholders and operationalise the above five principles to support VET stakeholders on the effective, secure and inclusive use of AI in VET development.

Establish national, education and VET-sector AI strategies with clear purpose and support

VET does not operate in isolation. Many OECD countries have initiated high-level AI strategies that shape the broader policy environment for AI in education and training. However, the evidence to date suggests that these strategies have not yet consistently “trickled down” into VET development: AI use in this area remains at an early stage and is often informal and exploratory, driven by individual practitioners or institutions rather than coherent system‑level strategies. Where AI use is not anchored in clear pedagogical and strategic objectives (including needs assessment, capacity building and links to quality assurance), it tends to remain small-scale, uneven and disconnected from core development workflows – potentially limiting scalability, sustained impact or stakeholder buy‑in.

Cross-country evidence on the gap between strategy and implementation shows this broader pattern. In the early stage of AI adoption in 2022, nearly all OECD countries (29 out of 30 countries or 97%) reported having a public-sector AI strategy, agenda or plan, but approaches to ensuring ethical management and use of algorithm varied: 16 countries (53%) relied on binding legal or regulatory requirements, 12 (40%) used softer instruments such as guidelines, standards or principles, and two (7%) – Latvia and Lithuania – reported no specific instruments (OECD, 2023[10]). By 2026, evidence from the OECD AI Observatory Index, which tracks effective implementation across 28 indicators and 138 countries, shows substantial variation in translating AI strategies into practice. Only a small group of countries score near the upper end of the scale – including Canada, Finland, Korea, Luxembourg, Norway, the United States, the United Kingdom and Switzerland (index value 0.55 or above in 2024). Most countries cluster at significantly lower levels, underscoring gaps in implementation capacity (OECD, 2026[11]).

In education and VET sector, national initiatives often focus on AI as a subject of learning or as a classroom-facing pedagogical tool than on transforming VET development processes, although these approaches can be complementary when strategically aligned. Country experience nevertheless illustrates that clear purpose, layered governance and targeted infrastructure and investment could help move beyond ad hoc experimentation.

• In England (United Kingdom), experimentation with AI-enabled tools for VET development, such as SkillsCompass and duties generator, has taken place within layered policy frameworks. At the strategic level, the UK National AI Strategy (2021) sets the overarching vision and principles for AI across government, complemented more recently by the AI Opportunities Action Plan (2025), which emphasises accelerated adoption and economic impact. Within the VET system, digital and data strategies, developed by the Institute for Apprenticeships and Technical Education (now Skills England), provided the operational basis for AI-related innovation, reliable and structured data, data protection and transparency in AI use. In addition, the Department for Science, Innovation and Technology (DSIT) has funded GBP 3 million to pilot “content store” of official assessments, curriculum guidance and teaching materials to train AI models aimed at reducing teacher workload (DfE, 2025[12]), alongside a GBP 1 million innovation competition (DfE, 2025[13]). The Department for Education also provided research insights on use cases for generative AI in education, drawing lessons from the development and testing of a proof-of-concept AI tool, regarding generating education-specific content with generative AI (DfE, 2024[14]).

• Estonia’s AI Leap Initiative aims to integrate AI into general and vocational education, starting in 2025‑2026, including teacher training and curriculum innovation, while other countries have established broader governance frameworks.

• Greece established a High-Level Advisory Committee on AI in 2023 to develop a national policy on AI opportunities and risks, with initial applications in VET and participation in EU-funded AI projects (OECD, 2025[1]).

• Lithuania’s Ministry of Education, Science and Sport, in co‑operation with social partners, developed national AI guidelines for education (2026). While not VET‑specific, the guidelines are relevant for VET schools, given the integration of general education curricula within many VET programmes. They provide an overarching framework for education institutions to define their own rules for AI use and to develop appropriate pedagogical and methodological approaches. Their preparation involved a broad coalition of stakeholders signalling growing system‑level engagement with AI governance. The guidelines are accompanied by enabling measures introduced in parallel, enabling access to advanced AI tools and nationwide training (SMSM, 2026[15]).

• The Netherlands positions itself as a leader in human-centric and trustworthy AI through initiatives such as the AiNed programme (a multi-year initiative to accelerate AI adoption, strengthen research capacity, and develop digital skills across sectors). under the National Growth Fund, and the Netherlands AI Coalition (NL AIC), which brings together over 400 stakeholders from industry, academia, civil organisations and government to foster innovation and ethical AI practices (OECD, 2025[16]). These strategies create enabling conditions for AI literacy. ethical competence and more co‑ordinated experimentation and development in VET.

• A high-level mapping shows that many countries have adopted national AI strategies or policies, some of which are (to be) followed by more concrete implementation. These include: Ireland’s National AI Strategy outlining the country’s vision to become a leader in using AI to benefit the economy and society; National AI Strategy or National Strategy on/for AI (Costa Rica; the Czech Republic (hereafter ‘Czechia’), Norway; Germany; France; Iceland; Japan); Policy for AI Development in Poland; National Policy on AI (Chile; Colombia); Strategic Vision for AI (Luxembourg); Pan-Canadian AI Strategy (Canada) (OECD.AI, 2025[17]).

In countries without clear national or sectoral AI strategies, AI use in VET development tends to be more fragmented. In Croatia, for example, stakeholders report isolated experimentation with generative AI to draft learning outcomes or structure curricula, but without a coherent national AI strategy or sustained institutional support but often with commercial interests (OECD, 2025[1]). Similar patterns are observed where smaller providers operate under resource constraints, reinforcing uneven adoption and limiting system‑level learning (OECD, 2025[1]). This experience underscores a broader policy gap: recognising AI’s potential is insufficient without co‑ordinated governance addressing privacy, transparency and stakeholder engagement, and support structures that guide AI use towards concrete challenges such as improving curriculum relevance, efficiency, accuracy, quality and labour market alignment.

Overall, the effectiveness of AI in VET development depends less on the tools themselves than on the clarity of purpose guiding their deployment and subsequent support. Establishing national, education and VET‑sector AI strategies with explicit objectives and assessed needs, defined use cases and aligned support mechanisms is therefore essential to move beyond fragmented experimentation and ensure that AI becomes a coherent driver of system‑level improvement rather than a collection of disconnected trials.

Manage risks by ensuring human-centred approach to AI use and AI-generated outputs

The benefits of AI hinge on managing risks (OECD, 2025[18]). AI is compelling yet challenging because of its opacity, algorithmic bias, hallucination. After all, AI still is not perfect (Maslej et al., 2025[19]). The most central and strategic direction to many risks associated with AI use is a human-centred approach to any processes involving AI use, relying on human judgement to manage its risks, especially core human skills such as critical thinking, reasoning and creative thinking, as safeguards against misuse and potential risks (OECD, 2026[4]). Many interviewed and surveyed stakeholders considered this human-centred approach as a solution to any potential risks of AI use and the evolving technology. It is acknowledged that the real danger lies when people fully trust what AI does, lower their guard and stop questioning.

VET curriculum and qualification design requires not only technical accuracy, but also contextual understanding, pedagogical coherence and alignment with labour market practices. Stakeholders across countries cautioned that AI outputs, if used without sufficient expert review, may fall short on these dimensions. In response, several countries are reinforcing human oversight mechanisms to address risk associated with AI use and concerns about the quality of AI-generated content. For example, Estonia has introduced pilot testing: new AI tools are tested on a small scale before wider application, allowing for quality control and refinement, and subject-matter experts review all the AI-generated content to ensure reliability and build trust in outcomes. Across survey responses, there was broad agreement that AI should function as a support tool rather than a substitute for professional judgement, with human validation remaining central to quality assurance.

The principle of human-centred, human validity and human-in-the‑loop is indispensable in the use of AI for developing VET for ensuring quality of AI-generated outputs, trustworthiness and responsibility. Expert judgment, contextual understanding and professional insight are still needed to guide AI-generated outputs. The risk is that AI may streamline tasks in the VET development at the expense of critical thinking of VET stakeholders. Effective use of AI in VET therefore hinges on maintaining human control – VET teachers’ and trainers’ experiential knowledge and industry partners’ hands-on experience must guide and interpret AI outputs rather than be overshadowed by them.

All countries participating in this study underline the central role of VET expert review and validation in ensuring the accuracy and compliance of AI-generated outputs. AI is viewed as a decision-support tool rather than a replacement for human judgement. In all countries that reported AI use in VET development, the final validation, their alignment with national frameworks and standards, and adaptation to pedagogical standards are the responsibility of experienced educators and curriculum designers. In Mexico, AI is integrated into various stages of curriculum development, but peer review and cross-validation by VET experts remain essential for ensuring quality and relevance.

Balance diverse perspectives and interests in using AI to improve VET development including local and institutional perspectives, while ensuring equal access to AI use

VET policymakers should balance diverse perspectives, interests and capacity in using AI to improve VET development. As the development of VET curricula and qualifications is a multi-stakeholder process involving multiple phases, tasks and institutions, stakeholders from government, industry and education are affected differently by their own AI-related policies and practices. Level of understanding, openness, skills towards AI differs across individuals, institutions and sectors, as do the interests they represent, the benefits they may gain and the risks they bear. For example, in Croatia, while AI is used in qualification and skills mapping by the Directorate for Labour Market and Employment and in the context of Application of Digital Technologies Based on AI in Education (BrAIn), the national VET agency and VET providers remain more reluctant to use AI in VET development. Some stakeholders including those in Germany and Switzerland see no particular need for AI use in this domain. As discussed in Chapter 3, the levels of AI use differ across VET stakeholders, and AI investments also vary across industries (McKinsey&Company, 2025[20]), which may affect the openness and capacity to adopt AI in VET development. Similarly, AI-supported tools may be more effective in fields with codified knowledge, standardised tasks and well-documented occupational standards, whereas programmes relying on tacit knowledge, practical judgement or interpersonal skills may be less suited to such applications.

Respecting this diversity also includes balancing the “innovation, changes and speed” associated with AI use, with “regulations, stability and predictability” on which VET systems depend. While AI has the potential to drive innovation in VET development by brining speed, efficiency and real-time labour market insight, VET systems require predictability and regulatory stability. Overly cautious or ambiguous regulatory signals may slow AI adoption or limit opportunities for efficient and creative solutions, while rapid uptake without safeguards may undermine quality. In addition, this divergence may risk a widening gap between innovative, market‑oriented providers and those operating under stricter regulatory oversight.

This tension is evident in countries with dual-market qualification systems, such as England and Korea, where government-regulated, publicly funded qualifications coexist with a growing segment of privately funded qualifications offered by commercial awarding organisations or private providers driven by employer demand. In such systems, government-regulated qualifications tend to prioritise risk mitigation, learner protection and public confidence, which may discourage providers to experiment with AI. In England, for example, the “compliant innovation” stance of the Office of Qualifications and Examinations Regulation – which recognises the potential for AI to support quality and innovation, yet prioritise fairness, validity and public confidence, and emphasise the risks of inappropriate or premature adoption of AI in high-stakes processes (Ofqual, 2024[21]) – is interpreted by some awarding organisations and VET providers as a “lack of public support” for providers’ AI adoption, although this relates primarily to assessment and grading rather than VET development more broadly. In Korea, private VET qualification and course providers are adopting AI more rapidly than public systems, which may pose challenges for quality assurance. Similar trade‑offs arise around data governance: in Germany, strong data protection and security remain priorities, yet strict restrictions may hinder the effective application of AI in VET qualification and curriculum development (OECD, 2025[1]).

Without clear and enabling frameworks for experimentation and innovation using AI, VET systems may under-utilise AI’s potential to support more responsive curricula, closer alignment with labour market needs and faster development of VET qualifications and courses for emerging occupations. Realising these benefits requires a balance between safeguarding quality and enabling innovation, combining clear and proportionate regulation and risk-based regulatory flexibility (OECD, 2025[22]) as well as active support for experimentation, capacity-building, and the development of sector-specific guidelines for AI use.

Co-create guidelines and build capacity for systemic and strategic AI use in VET development

Balancing the trade‑offs between efficiency gains and the risks of AI requires recognising VET’s distinctive features: heterogeneous programmes, occupations and providers, and consensus-based or collaborative governance involving diverse stakeholders. Hesitancy or reluctancy – observed in countries with strong social-partner traditions such as Germany and Switzerland – often reflects concerns about disrupting established processes or devaluing professional expertise. Addressing this resistance depends not only on technological availability but also on co-created guidelines, transparent communication, and sustained capacity building that embed AI responsibly and securely within established VET development processes.

In addition, transparent labelling of AI-generated content, clear agreements or guidelines on data use and confidentiality, formalised tool selection and deployment, and explicit human validation strengthened by targeted training can help build trust and mainline quality and ensure safe and effective use of AI in curriculum development and certification processes. In the Netherlands, some interviewed stakeholders called for transparent labelling of AI-generated content and the databases that can be used in AI systems to support VET development, as well as open communication about AI’s role and limitations.

Develop co-created guidelines and safeguards

Inspired by international initiatives, countries are increasingly institutionalising AI governance and developing tailored guidelines to balance innovation with legal and ethical safeguards, ideally involving stakeholders’ perspectives and allowing flexibility for VET stakeholders and rapid adaptation to technological change (see also Box 5.1):

• Netherlands: The Organisation for the Co‑operation between VET and the Labour Market (SBB) has established a Digital Ethics Commission to review and approve AI tools and projects, require risk assessments, and ensure compliance with legal, ethical and organisational values (Box 5.1). VET provider representatives also emphasise working with EdTech providers “on our terms” to safeguard autonomy and public values practiced in VET (OECD, 2025[1]).

• Finland: The Ministry of Education and Culture has issued guidance on AI use, transparency and human validation, shaping how VET providers experiment with AI in curriculum development (Finnish Ministry of Education and Culture, 2024[23]). AI guidelines (e.g. AI Guide for Teachers) emphasise teacher readiness, ethical use and strengthening AI-related competences to ensure safe and equitable adoption in education and training (OKM, 2024[24]).

• Mexico: COSFAC highlights the need for structured institutional guidance and both general and technical guidelines for AI use in curriculum development, including legal clarity and data protection before unpublished materials can be used with AI tools. Acknowledging the need, COSFAC is working to establish ethical guidelines and boundaries in using AI and data responsibly and securely (OECD, 2025[1]).

• Croatia: The BrAIn project (Application of Digital Technologies Based on Artificial Intelligence in Education) aims to foster a comprehensive understanding of AI among students, preparing them for future challenges and opportunities in a digitally-driven world, while ensuring that ethical considerations remain at the forefront of technological integration in education. The project also includes the development of digital educational content, training for educational staff, and research on the impact of digital technologies on students. A Committee for the Ethical Application of Digital Technologies and AI in Education has been established to monitor the integration of AI into the curriculum, ensuring that ethical considerations are addressed throughout the project’s implementation. This committee collaborates with entities such as the Laboratory for Ethical Aspects of Advanced Digital Technology at the University of Rijeka’s Center for Artificial Intelligence and Cybersecurity, which focusses on the ethical implications of AI and advanced digital technologies. Additionally, the Croatian Artificial Intelligence Association (CroAI) is involved in the BrAIn project, providing advisory support on curriculum development (OECD, 2025[1]).

• Nova Scotia (Canada): Nova Scotia Apprenticeship Agency is developing internal AI policies and best practice guidelines centred on transparency, quality assurance and stakeholder engagement (OECD, 2025[1]).

Box 5.1. Examples of relevant AI guidelines

Recommendation of the OECD Council on Artificial Intelligence (OECD AI Principles)

The OECD AI Principles – the first intergovernmental standard on AI – call to build human capacity for AI while promoting the stewardship of trustworthy AI. Adopted by the OECD Council at Ministerial level in May 2019, the Recommendation aims to foster innovation and trust by ensuring that AI development and use respect human rights and democratic values. It sets out five complementary values‑based principles – inclusive growth and well‑being; human‑centred values and fairness; transparency and explainability; robustness, security and safety; and accountability – with five policy recommendations related to research and innovation, digital ecosystems, enabling policy environments, human capacity and labour market transitions, and international co‑operation. It also encourages the development of metrics and evidence to monitor progress in implementation (OECD, 2019[3]).

Guidelines for effective, equitable and trustworthy use of generative AI in education (GenAI Guidelines)

The “Opportunities, guidelines and guardrails for effective, equitable and trustworthy use of generative AI in education” (OECD-Education International, 2023[8]) provide practical recommendations for the adoption of GenAI across education systems. The guidelines emphasise that AI should support the professional judgement and pedagogical role of teachers, while strengthening learning outcomes and inclusion. They recommend establishing clear governance frameworks for the use of GenAI, including transparency about how systems work, protection of student and teacher data, and mechanisms to identify and mitigate bias or inaccuracies. The guidelines also stress the importance of building educators’ and learners’ digital and AI literacy, ensuring equitable access to AI tools, and involving stakeholders in the design and implementation of AI solutions. Overall, they call for a balanced approach that harnesses the potential of generative AI to enhance teaching, learning and assessment while maintaining strong safeguards for ethics, accountability and educational integrity.

European Data Protection Board (EDPB)

Within the European Union, the Artificial Intelligence Act (European Parliament and Council of the European Union, 2024[6]) provides a binding regulatory reference for AI systems, including those used in education and training, where they are classified as high‑risk. It establishes requirements related to human oversight, responsibility allocation, risk management and documentation, reinforcing accountability and transparency. In parallel, the General Data Protection Regulation (GDPR) (European Parliament and Council of the European Union, 2016[5]) establishes sets the legal baseline for the lawful and secure processing of personal and sensitive data in AI‑supported processes. Guidance issued by the European Data Protection Board (EDPB) – most recently Opinion 28/2024 – further clarifies the application of GDPR to AI, emphasising lawful, fair, transparent and purpose‑specific data use, alongside obligations related to information provision, record‑keeping, risk and impact assessments, privacy‑preserving techniques, and robust security and governance arrangements. Complementing these instruments, the European Commission’s non‑binding Guidelines on prohibited AI practicesclarify which AI uses are deemed unacceptable in light of fundamental rights and European values, including certain practices relevant to educational contexts, and provide an additional reference when assessing permissible AI uses and appropriate safeguards. These regulatory and interpretative instruments provide a comprehensive reference framework for assessing permissible AI uses and appropriate safeguards in VET development.

The Netherlands

The Organisation for the Co‑operation between VET and the Labour Market (SBB) has established a Digital Ethics Commission within the organisation, comprising representatives from all departments, including privacy and data protection officers, tasked with evaluating and guiding the ethical and safe use of AI within the organisation. This Commission evaluates all AI use cases that could structurally impact organisational processes, ensuring that AI applications are ethical, lawful and safe. The Commission developed internal guidelines and digital ethical principles (not publicly available), inspired by international standards and guidelines such as OECD AI Principles and European Data Protection Board’s recommendations, and mapped them to SBB’s core values such as transparency and lawfulness. All significant AI projects must be reviewed by the Commission before deployment, with risks documented and mitigated.

The guidelines distinguish between “rules of play” for everyday AI use (e.g. no personal or confidential data, no use of high-risk external platforms) and broader ethical principles for project-level applications. The Commission’s work is seen as both a safeguard and an educational resource, fostering a culture of innovation. While SBB’s AI platform is only accessible to internal staff, the organisation recognises the importance of extending ethical oversight to partners and advocates for similar governance structures in other institutions, especially as AI use becomes more widespread.

Source: OECD (2025[25]), Recommendation of the Council on Artificial Innovation, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0449; OECD-Education International (2023[8]), Opportunities, Guidelines and Guardrails on Effective and Equitable Use of AI in Education, 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.

Strengthen targeted, role‑specific capacity building with structural support and continuous learning

AI literacy alone is insufficient; VET stakeholders require contextualised, task-specific skills and critical judgement to apply AI in qualification and curriculum development. While countries can rely on general AI awareness training (e.g. Germany) or teacher professional training (Switzerland), with limited sector-specific programmes tailored to VET development functions, more countries are beginning to address the challenge of limited AI literacy among VET stakeholders involved in VET development through targeted initiatives for the safe and effective use of AI.

• The Netherlands’ Organisation for the Co‑operation between VET and the Labour Market (SBB) provide targeted training and support for staff in using AI responsibly in qualification development, such as through summer schools for staff. For example, an external company provided prompt- engineering training, with participants using the SBB Chat platform (a secure GPT‑4 platform developed for drafting qualifications and automating technical checks) as a secure, hands-on environment to practise and apply the techniques during the training. The training was tailored to the specific daily tasks of each department, allowing staff to practice writing prompts relevant to their work, such as drafting legal requirements for qualifications and job and skill descriptions. Initially optional, the training became mandatory after strong uptake and positive feedback, embedding AI competence as a standard skill for all departments. The goal is to improve SBB staff skills in using AI tools effectively within their roles (OECD, 2025[1]).

• In England (United Kingdom), the Department for Education established community of practice to share learning, address concerns and build common understanding, complemented by AI awareness‑raising through trailblazer groups and route panels – employer‑led groups overseeing occupational standards and technical qualifications across major sectors. Dedicated AI leads were appointed across the 15 route panels to raise awareness and guide AI adoption. Beyond the VET system, AI training and upskilling opportunities are offered more broadly across government and the workforce, structured around AI user personas (leaders, professionals, workers and citizens) adapted from the Alan Turing Institute’s BridgeAI framework (InnovateUK-BridgeAI, 2024[26]) to tailor learning to different roles (OECD, 2025[1]).

• Estonia offers training and awareness-raising opportunities such as seminars and workshops to introduce AI use cases in qualification and curriculum development, aiming to build capacity among educators, developers and experts (OECD, 2025[1]).

• Lithuania and Croatia: EU Erasmus+ teacher-training initiatives have strengthened AI literacy and digital skills among VET teachers, with trained participants disseminating knowledge nationally and acting as change agents (OECD, 2025[1]).

• In Switzerland, when it comes to VET teacher training, VET teacher training colleges train future VET teachers how to use AI, including use for adjusting the national curricula and preparing school-level curricula and their lessons. For example, St. Gallen University of Teacher Education (PHSG) teach AI to future VET teachers using a platform, aprendo (https://aprendo.ch/). This platform was developed by the University in 2020 for digital continuing education programme and is open to all VET teachers and principals across Switzerland. It also uses AI for module development and offers practice‑oriented modules on effective use of AI, ICT application skills, digital professionalism and digital leadership. For example, it provides relevant modules to VET curriculum adjustment and implementation, such as AI & Digital Ethics, AI guidelines and standards for schools, AI tools for teaching, Researching with AI (OECD, 2025[1]).

• Regarding older adults’ AI literacy, OECD provides relevant, practical recommendations regarding digital skills for seniors for this persistent gap: such as simplifying complex concepts and providing age‑appropriate content and materials (OECD, 2025[27]).

Effective AI integration requires sustained institutional support, not one‑off training. Stakeholders frequently report enthusiasm but limited confidence in applying AI strategically to VET development. Continuous updating of skills, clarity about risks and unintended consequences, and reinforcement of critical thinking are essential.

• Switzerland illustrates systemic support structures: pedagogical ICT supporters (PICTS) guide teacher training and the strategic development of digital transformation in schools, while technical ICT supporters (TICTS) maintain technical support and infrastructure in schools. Such roles – encouraged by national initiatives such as “Digitale Schweiz” – help embed AI within broader digital transformation strategies. PICTS become increasingly common, especially in cantons that prioritise digital transformation, such as St. Gallen. They are often found in teacher training institutions and public VET schools. However, access to advanced tools and structured programmes remains uneven across countries and institutions, underscoring the need for collaborative platforms and equitable capacity-building strategies to ensure that AI adoption in VET development is strategic, inclusive and responsibly governed (OECD, 2025[1]).

• The Netherlands promotes collaborative platforms for secure AI use among VET providers to bridge institutional gap in equitable and effective AI adoption (OECD, 2026[28]).

Box 5.2. Capacity building models that could be drawn for AI literacy and digital skills for VET development

Improving the skills and competences of VET teachers and trainers in the age of AI (Taccle AI)

Taccle AI is an Erasmus+ funded project in five countries to provide initial training and continued professional development for VET teachers and trainers in AI. It seeks to support them in extending and adapting open curriculum models for incorporating AI in VET. The project has also developed a Massive Open Online Course in AI in education, free and open to all teachers and trainers in VET in Europe.

The policy recommendations that this project came up with include: (i) update VET curricula to include AI (ii) incorporate competences for AI into all initial training programmes for VET teachers and trainers (iii) encourage and support the development, searchability and sharing of open educational resources for AI in VET (iv) encourage and support the development of online programmes of Continuing Professional Development for AI in VET, (v) support collaboration between industries, VET schools and training centres.

A discussion from the project is the emergence of the “augmented operator” – a worker who integrates AI-assisted decision making with a systemic understanding of production processes – driven by the rise of AI in technological work processes. This shift requires VET curricula to go beyond basic ICT skills and incorporate interdisciplinary, work-process-oriented training. Augmented operators must be equipped to manage smart technological chains, big data, networking, and information security while considering economic, ergonomic, and environmental implications. Traditional competence‑based VET models may struggle with this holistic approach, necessitating stronger work-based learning, integration of academic knowledge, and collaboration between VET and higher education providers. As a result, qualifications for augmented operators may be positioned at Level 5‑8 of qualifications frameworks, bridging vocational and higher education.

Data Literacy Initiative (DaLI) model for Higher Education and Research

The DaLI model at TH Köln can provide a structured foundation for developing AI literacy and digital skills in education by organising competencies across seven interconnected areas that mirror the full lifecycle of data and AI use: Establish Data Culture, Provide Data, Manage Data, Analyse Data, Evaluate Data, Interpret Data, and Publish Data. Each competence area is further broken down into specific dimensions and progression levels, enabling a systematic approach to curriculum design, teaching, and assessment. This tiered design supports VET institutions in building capacity gradually – from fostering shared organisational understanding of data and AI principles, through practical data handling and analysis skills, to the critical evaluation, interpretation, and communication of insights. As such, the model can serve as a blueprint for designing targeted learning pathways and upskilling strategies tailored to different roles involved in VET development (Echtenbruck et al., 2025[29]).

Guide on system-level AI adoption based on AI maturity

The European Commission’s Joint Research Centre (JRC) has developed a governance‑ and competence‑based framework to guide system-level AI adoption in public sector organisations, including education and training. Rather than prescribing specific technologies, it defines the institutional conditions needed for sustainable and effective AI integration at scale (e.g. a bespoke AI system embedded in a VET system or a shared platform across institutions). The model emphasises that AI maturity depends as much on governance and organisational capacity as on technology, identifying three core competence areas – technical, managerial, and legal-ethical – supported by clear strategic direction, accountability structures, human oversight, risk management, data management and continuous monitoring. Applied to VET, a JRC-aligned approach would involve controlled data access, integration with qualification frameworks, and clearly defined roles for curriculum experts, social partners and public authorities, with AI supporting tasks like skills analysis or curriculum comparison while final decisions remain under human control. Overall, the framework provides a reference point for moving from fragmented experimentation to co‑ordinated, trustworthy system-level AI adoption (Medaglia, R., P. Mikalef and L. Tangi, 2024[30]).

AI maturity model and strategic framework and toolkit, by Joint Information Systems Committee

Jisc is a registered charity and the key digital, data and technology partner for tertiary and further education (FE) and research across the United Kingdom, supporting all FE colleges and universities. The Jisc AI maturity model in England provides a system-level enabling framework for AI adoption in tertiary education, including FE and skills providers, without mandating a single national AI system. It offers a shared reference structure that helps institutions move from ad hoc experimentation to more embedded and governed AI use by defining common maturity stages, establishing shared principles and language, and linking technology choices to organisational culture, skills, governance and data readiness. While implementation decisions remain decentralised, the framework supports cross-institutional learning and strategic alignment. Unlike a centrally embedded AI platform, Jisc does not impose specific tools for curriculum or qualification development; instead, it provides guidance, toolkits and governance support, enabling co‑ordinated but locally configured AI adoption within a common strategic framework.

Source: Attwell et al. (2021[31]), Artificial Intelligence & Vocational Education and Training, https://taccleai.eu/wp-content/uploads/2021/12/TaccleAI_Recommendations_UK_compressed.pdf; Echtenbruck et al. (2025[29]), A Data Literacy Competence Model for Higher Education and Research, https://arxiv.org/abs/2504.15690; Jisc (2024[32]), New toolkit for colleges and universities sets pathway for effective adoption of AI, www.jisc.ac.uk/news/all/new-toolkit-for-colleges-and-universities-sets-pathway-for-effective-adoption-of-ai; Webb (2024[33]), Our AI in Education Maturity Model – an update for 2024, https://nationalcentreforai.jiscinvolve.org/wp/2024/03/08/our-ai-in-education-maturity-model-an-update-for-2024; Jisc (2024[34]), New toolkit for colleges and universities sets pathway for effective adoption of AI, www.jisc.ac.uk/news/all/new-toolkit-for-colleges-and-universities-sets-pathway-for-effective-adoption-of-ai.

Improve data availability, quality, interoperability and usability for effective and secure AI use

Many countries already maintain rich and structured databases on VET curricula, qualifications and occupational standards. If made machine‑readable and interoperable, these resources could substantially strengthen AI-supported curriculum design and qualification development, while improving alignment with labour market needs. The better structured and standardised the data, the more scalable and reliable AI applications become. Machine‑readable learning outcomes, for example, enable AI to map qualifications to occupational standards, track revisions across curriculum versions, and process large volumes of textual data using natural language processing to support a common language for describing skills – facilitating both local-national coherence and cross-border recognition and mobility.

However, realising this potential depends on addressing well-known AI risks (limited transparency of models, algorithmic bias), as well as VET-specific data issues such as heterogeneous qualification formats, complex educational language, uneven data quality, contextual sensitivity, and limited integration of AI outputs into formal qualification design processes (Cedefop & UNESCO-UNEVOC, 2025[35]). Fragmented databases and inconsistent model use further reinforce the need for robust data governance frameworks and sustained human oversight.

OECD evidence shows that structured and ethically sound data acquisition frameworks are central to building effective and trustworthy AI systems (OECD, 2025[36]). Across 38 jurisdictions, organisations with structured, ethical and legally compliant data governance protocols achieve 15‑22% better model performance while reducing bias and regulatory risks. Approaches such as federated learning, synthetic data and cross-institutional data sharing enable AI training without compromising privacy or proprietary constraints. Clear data stewardship and standardised metadata also speed up deployment – by up to 18% – and strengthen stakeholder trust, offering important lessons for AI use in VET development (OECD, 2025[36]). Based on this evidence, effective frameworks rest on three interlinked pillars for effective AI data collection, trustworthy and high-performing AI applications – integration of these pillars enhances predictive accuracy, mitigates systemic bias, and supports transparency in AI development (OECD, 2025[36]):

• Ethical and legal compliance, encompassing privacy, consent, intellectual property;

• Technical robustness, ensuring data quality, diversity, representativeness for reliable training outcomes);

• Institutional co‑ordination, fostering interoperable frameworks across public, private and research actors).

Given the sensitivity of educational and labour market data used in VET development, countries are combining improved data access with stronger safeguards such as data protection protocols to strengthen data protection and secure AI infrastructures. Some countries are investing in secure, privacy-compliant AI solutions and formalising the selection and evaluation of AI tools.

• Germany: The Competence Compass from Institut für Arbeitsmarkt- und Berufsforschung (IAB), funded by the Federal Ministry of Labour and Social Affairs (BMAS) under NextGenerationEU, develops structured datasets from job advertisements to extract standardised information on skills, tasks and qualifications from unstructured texts, using natural language processing and improving data quality and usability for AI‑supported analysis. This enables granular tracking of skill trends – including AI-related skills – and supports evidence‑based workforce planning and curriculum alignment (Stops et al., 2025[37]).

• Croatia: Algebra University runs AI models on-premises, creates synthetic datasets for testing and improving curriculum-related AI models without exposing sensitive and private information, and tailor outputs to institutional needs. It uses large language models for drafting and reviewing content and advanced visualisation tools (e.g. Power BI, Tableau) for decision support, while keeping data within secure internal systems. The BrAIn initiative also stresses ethical oversight and expert involvement in curriculum-related AI use.

• Netherlands: The Organisation for the Co‑operation between VET and the Labour Market (SBB) and FME restrict the sharing of sensitive data with external AI platforms. Their internal policies prevent exposure of confidential information to external AI platforms or cloud environments – e.g. sensitive company and salary data remain within secure internal systems to avoid data leaks and misuse. For industry and associations, this is not only for data protection, but also not to lose commercial advantage and control over valuable data. Stakeholders warn against uploading student data or confidential materials into public AI tools without clear licensing and privacy safeguards.

• In Nova Scotia (Canada), public curriculum and occupational standards are considered suitable for AI processing, while exam development and commercial curricula require additional safeguards, reflecting a differentiated, risk-based approach.

• Estonia: Efforts focus on improving interoperability and reducing fragmentation across databases. Estonia collaborates with private initiatives such as LAiSER to enhance labour market data integration and leverages AI to streamline data systems. AI data and system security issue is not VET’s own; national level strategy and measures ensure the secure and effective use of AI in VET. For example, the secure data exchange platform, X-tee, enables public and private organisations to share and access data automatically and securely, supporting privacy-compliant AI use (Box 5.3).

Box 5.3. What enables and ensures secure use of AI in VET?

Estonia’s X-tee

X-tee acts as a foundational digital infrastructure in Estonia, securely connecting databases and information systems across sectors (e.g. government, health, education, finance). It allows different IT systems to communicate, regardless of their technology or architecture. There is no central database; instead, X-tee connects existing databases, allowing data to remain with its owner. All data exchanges are encrypted, logged, and monitored. Only authorised parties can access the data, and all actions are traceable. The platform supports thousands of organisations and millions of transactions per day. It also automates data sharing, reducing paperwork and manual processes. Data requests and responses are handled via secure, standardised interfaces. Every transaction is encrypted and logged for transparency and accountability.

It ensures data integrity and privacy, complying with GDPR and Estonian standards, enabling new digital services and cross-sector collaboration (e.g. e‑health, e‑tax or e‑school).

X-tee’s technology has been adopted by several other countries (e.g. Finland, Iceland, Faroe Islands, and more), supporting cross-border digital services.

Source: Estonia Information System Authority (2024[38]), Data exchange layer X-tee, www.ria.ee/en/state-information-system/data-exchange-platforms/data-exchange-layer-x-tee.

5.3 Next steps

AI offers a timely opportunity to strengthen the responsiveness, efficiency and evidence base of VET curriculum and qualification development. As this report has shown, AI can support long‑standing challenges in VET systems by improving the synthesis of labour market information, accelerating drafting and revision processes, and complementing multi‑stakeholder consultation. Used responsibly, AI can help VET systems adapt more quickly to technological change, digitalisation and the green transition, while preserving, or improving, the rigour, relevance and legitimacy that underpin trust in VET curricula and qualifications.

At the same time, the report highlights that AI use in VET development remains at an early and uneven stage, characterised by fragmented experimentation, limited institutional capacity and unaddressed governance questions. Risks related to data protection, quality assurance, accountability and the possible weakening of collaborative decision‑making are real and must be actively managed. These challenges are not purely technical: they reflect institutional arrangements, organisational practices and professional cultures within VET systems. Ensuring that AI strengthens rather than undermines these foundations requires clear purpose, human‑centred design, transparency and strong safeguards.

Looking ahead, moving from isolated pilots towards more strategic and systemic use of AI across VET system will require coherent policy direction, VET‑specific guidelines, continued capacity and evidence building, and investment in high‑quality, interoperable data infrastructures. By grounding AI use in shared, but VET-specific, principles and steering the use through appropriate policy measures as suggested in the form of policy considerations in this report, countries can harness AI’s potential while maintaining quality, trust, equity and accountability. In doing so, AI can become an excellent supportive instrument – rather than a disruptive force – in building more agile, inclusive and future‑ready VET systems.

The overarching objective of the OECD’s work in this area is to build an evidence‑based, system‑wide understanding of how AI can support the full VET cycle – from curriculum and qualification design to teaching and learning, assessment and the monitoring of learner outcomes. Building on this work, further work will be needed to examine how AI can augment VET delivery, assessment and the monitoring of learner outcomes across the full VET cycle – an opportunity to address VET’s unique challenges in these policy areas from a holistic perspective. This points to several priority avenues for future research and policy dialogue.

References

[31] Attwell, G. et al. (2021), Intelligence & Vocational Education and Training, https://taccleai.eu/wp-content/uploads/2021/12/TaccleAI_Recommendations_UK_compressed.pdf.

[35] Cedefop & UNESCO-UNEVOC (2025), Meeting skill needs for the green transition: Skills anticipation and VET for a greener future, Publications Office of the European Union, Luxembourg, https://doi.org/10.2801/6833866.

[13] DfE (2025), AI teacher tools set to break down barriers to opportunity, https://www.gov.uk/government/news/ai-teacher-tools-set-to-break-down-barriers-to-opportunity.

[12] DfE (2025), Generative artificial intelligence (AI) in education, https://www.gov.uk/government/publications/generative-artificial-intelligence-in-education/generative-artificial-intelligence-ai-in-education?pStoreID=EP11678.

[14] DfE (2024), Use Cases for Generative AI in Education, https://assets.publishing.service.gov.uk/media/671108a18a62ffa8df77b2bf/Use_Cases_for_Generative_AI_in_Education_-_Technical_report_October_2024.pdf.

[29] Echtenbruck, M. et al. (2025), A Data Literacy Competence Model for Higher Education and Research, https://arxiv.org/abs/2504.15690.

[38] Estonia Information System Authority (2024), Data exchange layer X-tee, https://www.ria.ee/en/state-information-system/data-exchange-platforms/data-exchange-layer-x-tee.

[6] 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), https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng.

[5] European Parliament and Council of the European Union (2016), Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data (General Data Protection Regulation), https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng.

[23] Finnish Ministry of Education and Culture (2024), Recommendations for artificial intelligence, https://okm.fi/en/project?tunnus=OKM021:00/2024.

[7] G7 (2024), G7 Toolkit for Artificial Intelligence in the Public Sector, https://www.oecd.org/en/publications/g7-toolkit-for-artificial-intelligence-in-the-public-sector_421c1244-en.html.

[26] InnovateUK-BridgeAI (2024), AI Skills for Business Competency Framework, https://zenodo.org/records/11092677.

[32] Jisc (2024), New toolkit for colleges and universities sets pathway for effective adoption of AI, https://www.jisc.ac.uk/news/all/new-toolkit-for-colleges-and-universities-sets-pathway-for-effective-adoption-of-ai.

[34] Jisc (2024), New toolkit for colleges and universities sets pathway for effective adoption of AI, https://www.jisc.ac.uk/news/all/new-toolkit-for-colleges-and-universities-sets-pathway-for-effective-adoption-of-ai.

[19] Maslej, N. et al. (2025), Artificial Intelligence Index Report 2025, AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, Stanford, CA, https://hai.stanford.edu/assets/files/hai_ai_index_report_2025.pdf.

[20] McKinsey&Company (2025), Superagency in the workplace: Empowering people to unlock AI’s full potential, https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work.

[30] Medaglia, R., P. Mikalef and L. Tangi (2024), Competences and governance practices for artificial intelligence in the public sector, Publications Office of the European Union, Luxembourg, https://publications.jrc.ec.europa.eu/repository/handle/JRC138702.

[9] 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.

[4] 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.

[28] 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.

[11] OECD (2026), The OECD.AI Index, OECD Publishing, Paris, https://doi.org/10.1787/32c01014-en.

[16] OECD (2025), AI Coalition of the Netherlands, https://oecd.ai/en/dashboards/policy-initiatives/ai-coalition-of-the-netherlands-3341.

[22] OECD (2025), Better Regulation Practices across the European Union 2025, OECD Publishing, Paris, https://doi.org/10.1787/6f007516-en.

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

[27] OECD (2025), “Digital skills for seniors: Key steps for effective training programmes”, OECD Policy Briefs, No. 35, OECD Publishing, Paris, https://doi.org/10.1787/9edfa7ef-en.

[18] OECD (2025), Governing with Artificial Intelligence: The State of Play and Way Forward in Core Government Functions, OECD Publishing, Paris, https://doi.org/10.1787/795de142-en.

[36] OECD (2025), “Mapping relevant data collection mechanisms for AI training”, OECD Artificial Intelligence Papers, No. 48, OECD Publishing, Paris, https://doi.org/10.1787/3264cd4c-en.

[25] OECD (2025), Recommendation of the Council on Artificial Intelligence, OECD Publishing, Paris, https://legalinstruments.oecd.org/en/instruments/oecd-legal-0449.

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

[3] OECD (2019), OECD AI Principles, revised 2023 and 2024, OECD Publishing, Paris, https://www.oecd.org/en/topics/ai-principles.html.

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

[8] OECD-Education International (2023), Opportunities, Guidelines and Guardrails on Effective and Equitable Use of AI in Education, OECD Publishing, Paris, 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.

[21] Ofqual (2024), Ofqual’s approach to regulating the use of artificial intelligence in the qualifications sector, https://www.gov.uk/government/publications/ofquals-approach-to-regulating-the-use-of-artificial-intelligence-in-the-qualifications-sector/ofquals-approach-to-regulating-the-use-of-artificial-intelligence-in-the-qualifications-sector.

[24] OKM (2024), Recommendations for artificial intelligence, https://okm.fi/en/project?tunnus=OKM021:00/2024.

[15] SMSM (2026), Parengtos dirbtinio intelekto naudojimo gairės – tai padės kiekvienai mokyklai nusistatyti savo DI naudojimo taisykles, rengiama didaktika ir metodai, https://smsm.lrv.lt/lt/naujienos-1/pranesimai-ziniasklaidai-1/parengtos-dirbtinio-intelekto-naudojimo-gaires-tai-pades-kiekvienai-mokyklai-nusistatyti-savo-di-naudojimo-taisykles-rengiama-didaktika-ir-metodai-9NiE/.

[37] Stops, M. et al. (2025), “IAB Competence Compass”: Provision of job description data as a structured dataset, https://doi.org/10.48720/IAB.FB.2501.

[2] Varsik, S. and L. Vosberg (2024), “The potential impact of Artificial Intelligence on equity and inclusion in education”, OECD Artificial Intelligence Papers, No. 23, OECD Publishing, Paris, https://doi.org/10.1787/15df715b-en.

[33] Webb, M. (2024), Our AI in Education Maturity Model – an update for 2024, https://nationalcentreforai.jiscinvolve.org/wp/2024/03/08/our-ai-in-education-maturity-model-an-update-for-2024/.