AI to Support Neurodivergent Learners in Vocational Education and Training

5. Policy guidelines

This chapter identifies policy guidelines (summarised in Box 5.1) to help governments seize the potential of AI and other advanced technologies to support neurodivergent learners in VET (Chapter 2) while overcoming barriers (Chapter 3) and addressing the associated risks (Chapter 4). The guidelines are accompanied by insights and practices captured during stakeholder interviews, desk research and a workshop bringing together additional stakeholders engaged in VET and disability policy.

Box 5.1. Policy guidelines for using AI and other advanced technologies to support neurodivergent VET learners

Apply existing frameworks governing accessibility, disability rights, AI, data privacy etc., by for instance:

• Reviewing existing frameworks in the context of the latest advances in AI and other advanced technologies

• Providing guidance on how to interpret existing legislation in the context of the latest advances in AI and other advanced technologies

• Developing and promoting technical accessibility standards for AI-enabled tools used in VET

• Including neurodivergent voices in discussions about regulation and governance of AI and other advanced technologies

Prepare and support VET teachers to use AI and other advanced technologies to support neurodivergent learners:

• Including training on assistive technologies, including AI and other advanced technologies, in initial teacher education

• Leveraging continuing professional learning to keep in-service VET teachers’ skills up to date and ensuring that instruction adapts to the needs of the labour market

• Promoting a culture of innovation among teachers and encouraging VET institutions to do the same

Help neurodivergent learners, VET institutions and employers to navigate the many AI and other advanced technologies available, by for instance:

• Providing information and guidance to help learners, teachers and employers keep pace with new tools, and to choose the right tools for their circumstances

• Encouraging VET institutions to provide clear guidelines on use of AI and other advanced technologies to learners on how they can use generative AI and writing assistance tools without breaching plagiarism rules

• Facilitating partnerships among VET institutions and centralised evaluations and procurement processes to select tools

• Advancing research on the use of AI and other advanced technologies to support neurodivergent learners in VET

Fund assistive technologies and support the assistive technology ecosystem, by for instance:

• Using government funding to support socially beneficial innovations and to bridge education and employment gaps.

• Supporting innovation by funding pilots

• Fostering healthy assistive tech ecosystems at regional, national and international levels

Encourage developers to improve the accessibility of AI and other advanced technologies and to better align them with the needs of neurodivergent learners, by for instance:

• Encouraging developers to engage neurodivergent learners (plus their parents, teachers and support professionals) in the development and testing of tools intended to serve their needs

• Encouraging the companies developing these tools to train developers about accessibility and bias

Use AI and other advanced technologies to help achieve more responsive, inclusive and innovative VET systems, by for instance:

• Making VET more responsive with data and AI

• Keeping VET relevant amid rapid technological change through employer engagement and work-based learning

• Using AI and advanced technology to broaden accessibility and inclusion in VET

Apply existing frameworks governing accessibility, disability rights, AI, data privacy etc.

Existing frameworks, ranging from legislation to “soft law” (which sets standards but is not enforceable), provide a starting point for using AI and other advanced technologies to support neurodivergent VET learners, while addressing the risks. Some interviewees expressed appreciation for EU legislation, such as the EU AI Act, GDPR and the European Accessibility Act. Colm McNamee described the European Union as playing a leading role in regulating issues such as ethical, data and environmental concerns associated with AI through soft power and standard setting. Veronika Kaska saw the EU AI Act as important because it sets minimum standards for AI-enabled tools used in education, while Kellie Mote mentioned that the European Accessibility Act had been successful in bringing attention to digital accessibility. Governments can apply existing governance frameworks by:

Reviewing existing frameworks in the context of the latest advances in AI and other advanced technologies. Rapid technological developments raise the possibility that gaps could arise or that regulatory frameworks could unintentionally block useful innovations (e.g. in highly personalised learning or emotion recognition tools to support learners with ASD).

Providing guidance on how to interpret existing legislation in the context of the latest advances in AI and other advanced technologies. Developers and deployers (including employers and VET institutions) of these tools could benefit from guidance on how to interpret and comply with existing legislation, such as the EU AI Act and GDPR. Yonah Welker mentioned an audit tool, currently being developed by a startup, which aims to help SMEs working on assistive and medical technologies create documentation to demonstrate compliance.

Developing and promoting technical accessibility standards for AI-enabled tools used in VET. Just as web accessibility standards (e.g. Web Content Accessibility Guidelines (WCAG)) explain how to make web content more accessible to people with disability, Sandra Fomotškin and David Banes suggested that technical accessibility standards could be developed for AI-enabled tools in VET.

Including neurodivergent voices in discussions about regulation and governance of AI and other advanced technologies. David Banes made the point that if policymakers legislate on the basis of the risk-benefit ratio (e.g. regarding privacy and security risks) for the neurotypical population, then neurodivergent people could be denied access to tools that are genuinely helpful to them.

Prepare and support VET teachers to use AI and other advanced technologies to support neurodivergent learners

Interviewees highlighted a need to prepare and support teachers (including trainers) to use these tools to support neurodivergent learners in VET. Not only would this enable teachers to use the tools pedagogically and effectively, but it would also help them to choose the most suitable tools, to know when to use them and when not to use them, to address risks and to transmit their knowledge to students. Having teachers with up-to-date knowledge and skills is a key component of responsive VET systems, according to an OECD report on building future‑ready VET systems (OECD, 2023[31]). As well as preparing and supporting VET teachers to respond to neurodiversity more generally, VET systems should equip teachers and potentially align pedagogy by integrating assistive and advanced technologies into initial teacher education and continuing professional learning, with practical, classroom-ready guidance, model lesson plans and safe‑use protocols, including how to scaffold use for neurodivergent learners and adapt assessment where needed (OECD, 2025[52]; OECD, 2023[31]). Governments can equip VET teachers with the necessary skills and knowledge by: Including training on assistive technologies, including AI and other advanced technologies, in initial education for teachers and support professionals. Rohan Slaughter outlined the different skill levels required for different profiles: (1) basic awareness of assistive technology tools and how they work for all teachers, (2) higher-level training for SEN teachers, occupational therapists and speech and language therapists, (3) specialist programmes to develop expert assistive technology practitioners for SEN and further education institutions. According to previous OECD research (2023[31]), the value of simulators, VR and other digital tools in classrooms and workplaces is realised when they are tied to clear learning goals and occupational standards. Training would need to account for the fact that prospective VET teachers and in-company trainers come from varied backgrounds, including from industry, and come equipped with varied levels of digital skills and AI literacy. Susan Scott-Parker suggested that teacher certification processes could include training on assistive technologies.

Leveraging continuing professional learning to keep in-service VET teachers’ skills up to date and ensure that instruction adapts to the needs of the labour market. In Estonia, training on AI use for (primarily non-VET) teachers has begun via AI Leap, with beginner-level workshops led by educational technologists, according to Sandra Fomotškin, who noted plans to extend the programme to VET teachers by 2026. Some VET institutions already provide training on AI and other advanced technologies to teachers. Luovi Vocational College offers monthly online training sessions, an internal knowledge hub and digital events open to external students and teachers.

Promoting a culture of innovation among teachers and encouraging VET institutions to do the same. Lorenzo Desideri and Thomas Köhler called for teachers to be given more freedom to explore the use of AI and advanced technologies. Piia Jokelainen described how Finland’s Ministry of Education recently issued AI guidelines, supporting the use of AI in education in principle. Collaboration platforms are also enabling more coherent delivery across learning sites, for example through shared logs, multimedia evidence and joint assessment spaces. This supports workplace‑embedded projects and faster iteration of teaching materials. Kevin Gonyop Kim highlighted that projection-based AR and shared 3D environments are often more workable than individual headsets for group learning and that AI can lower barriers by auto-generating VR or AR content and supporting teachers through conversational build tools. Open 3D libraries allow teachers to reuse high-quality models in practical subjects, which helps extend innovation beyond academic domains. Policy should also address disparities in digital capacity across VET institutions to ensure that innovation benefits all learners, not only those in well-resourced centres.

Help neurodivergent learners, VET institutions and employers to navigate the many AI and other advanced technologies available

Interviewees described how governments could help neurodivergent learners, VET institutions and employers navigate the many AI and other advanced technologies available by providing information, facilitating partnerships between VET institutions and furthering research on the potential of AI and other advanced technologies to support neurodivergent learners in VET. Governments can help these groups navigate the many AI and other advanced technologies available by:

Providing information and guidance to help learners, teachers and employers keep pace with new tools, and to choose the right tools for their circumstances. Neil Miliken highlighted the need for reliable catalogues of tools that are regularly updated. According to Susan Scott-Parker, a “buyer’s guide” is urgently needed – not just a list of tools, but guidance on how to evaluate, select and justify purchasing decisions. Geena Vabulas called for stronger government involvement in filtering, curating and disseminating trustworthy information about available tools. Kevin Polley (Digisprong) reported that the Flanders Government has provided a collaborative document outlining seven principles, including transparency, data privacy and cybersecurity, for schools to use to assess tools. In many countries, sectoral bodies or employer associations play a key role in co‑ordinating VET provision and could also help evaluate and disseminate AI-based assistive technologies within their occupational domains.

Encouraging VET institutions to provide clear guidelines on use of AI and other advanced technologies to learners on how they can use generative AI and writing assistance tools without breaching plagiarism rules. In the view of Yuwei Lin (Senior Lecturer, University of Roehampton), the most important aspect of this is providing practical guidance to learners on how to disclose when and how they have used generative AI. David Voss reported that his institution had recently produced a toolkit to provide transparency, including on the issue of plagiarism, for all students and teachers around the use of generative AI. Helen Nicholson-Benn described how her organisation Jisc is producing accessible, non-technical guides to help learners and their institutions to understand the tools, their features and how they might interact with existing rules.

Facilitating partnerships among VET institutions and centralised evaluations and procurement processes to select tools. Interviewees noted the advantages in terms of efficiencies and information sharing. Piia Jokelainen made the point that centralised evaluation of tools would be far less resource intensive than the current norm where each school is responsible for managing its own selection and implementation process. Her VET institution already participates in a national network of five special needs VET providers. Susan Scott-Parker suggested that government-convened procurement consortia across schools and VET providers could enable purchases at scale, cutting prices and widening access. In the United Kingdom, Jisc sets up licensing partnerships with providers of assistive technologies, enabling supports its members, further and higher education institutions, to benefit from lower prices. It also acts as an impartial advisor, publishing blog posts providing guidance on how to pick software right for each institution.

Advancing research on the use of AI and other advanced technologies to support neurodivergent learners in VET. According to Yonah Welker, research can help inform the public about the potential of AI, while correcting misconceptions and ultimately informing conversations about how to regulate AI used to support neurodivergent learners. This report highlights a number of areas where more research is needed, such as the long-term effects of the use of AI on skill acquisition and on social-emotional development for neurodivergent (and neurotypical) learners. Kevin Polley (Digisprong) noted that the Flemish Government has launched a research call to assess the actual use of AI in schools, how it is being applied, and in which contexts. The European Vocational Training Association (EVTA) has established a new project, Recognise, with the aim to integrate AI tools into VET. The project will comprise a research phase on educational tools within VET, training for teachers and career guidance activities. Cristina AnaMaria Costescu called for government to support rigorous testing on available tools to enable neurodivergent learners to choose the best tools based on solid evidence. In the United Kingdom, the Westminster Department for Education commissioned computing experts at the University of Dundee (including interviewee Rohan Slaughter) to draft a competency framework for staff in special schools and colleges using digital assistive technologies (Slaughter and Griffiths, 2025[53]).

Fund assistive technologies and support the assistive technology ecosystem

Interviewees agreed that government has a role to play in funding assistive technologies, supporting socially beneficial innovation and fostering healthy assistive technology ecosystems, especially for tools used in VET and other education pathways. Governments can support the assistive technology ecosystem by:

Using government funding to support socially beneficial innovations and to bridge education and employment gaps. Many interviewees called for higher funding to bridge education and employment gaps, since access to AI and advanced technologies for VET institutions, employers and individuals often relies on government funding. Interviewees made the point that government and public institutions should take an interest in financing AI and other advanced technologies that are accessible, equitable and proven to work, especially in VET, given the important economic and social role that it plays by engaging diverse learners, equipping them with practical and job-specific skills, and facilitating transitions from school to work. Clayton Lewis suggested that government funding could sustain tools that are effective but have limited commercial potential (e.g. due to the limited number of potential users), a point also made in previous OECD research (Touzet, 2023[46])). Pavan Konanur called for more public-private partnerships, highlighting the advantages in combining private sector technical expertise with public sector expertise on health and special education.

Supporting innovation by funding pilots. Interviewees saw government funding for pilots as a way to strengthen the evidence of assistive technologies’ impacts, enabling the developers of successful pilots to attract further funding and to scale across VET. However, they suggested different approaches. Robert McLaren (Policy Connect) called for government to focus on large‑scale pilots which could potentially show a large return on investment and thus lead to widespread integration of worthy tools across VET. Motti Sigel (MassChallenge) suggested that governments should focus on small-scale pilots, encouraging schools to experiment with new tools and rewarding measurable success, while minimising risk. Experimentation at VET institution level, which can put tools to the test in real learning environments, may be easier in countries and regions where institutions have more independence. Kevin Polley described a Flemish initiative, the Smart Education at Schools (SEATS) programme, whereby schools co-develop tools with tech partners based on local needs and aligned with real classroom challenges. Tools developed must be shared openly with other schools on an online platform, and tech companies can buy the solution to be further developed commercially (in which case the basic version must remain available for schools).

Fostering healthy assistive tech ecosystems at regional, national and international levels. Governments can play a role in fostering healthy assistive tech ecosystems, comprising: a diverse range of developers (labs, universities, start-ups, NGOs, social enterprises, large tech companies); accelerators and other development programmes to fund and scale initiatives; and policymakers who establish frameworks and regulations to ensure inclusion and safety. Yonah Welker spoke of the importance of decentralised and democratised systems where a variety of labs, testing environments and developers produce different assistive technologies – and avoiding the case where one large tech company monopolises the market. In his view, EU funding programmes work well in this respect because they allow several labs and/or developers to collaborate on one technology. He also made the point that healthy national or regional assistive tech ecosystems would further help produce tools that are suited to local conditions, languages, VET systems and labour markets. Ricardo Rosas shared the view that international governance (e.g. by UNESCO) could help avoid fragmented efforts in developing AI tools, especially where the target populations are relatively small. He suggested convening a global consortium to set open technical and data standards, build shared training datasets, and publish validation and procurement guidance. His concrete example was AI sign-language avatars: several teams are developing similar tools in parallel, but without common standards and pooled investment these projects are unlikely to converge or scale across countries and languages.

Encourage developers to improve the accessibility of AI and other advanced technologies and to better align them with the needs of neurodivergent learners

Governments can encourage companies developing these tools to actively engage neurodivergent learners in the process and to train developers about accessibility and bias, to address the concern that many tools currently on the market do not align with the real needs of neurodivergent learners in VET. Governments can do this by:

Encouraging developers to engage neurodivergent learners (plus their parents, teachers and support professionals) in the development and testing of tools intended to serve their needs. The needs and voice of the intended user need to be built in from the beginning rather than added as an afterthought, according to Kellie Mote and Colm McNamee. Co-design (i.e. collaborating with neurodivergent individuals in the design and development process) could help tools align better with needs. Kellie Mote highlighted the need to employ neurodivergent people in the tech industry and/or to pay them for their contributions to the tools they help develop and test. Cristina Anamaria Costescu (Babeș-Bolyai University) described co-design as crucial to all tools she has helped develop. Hiren Shukla described how Microsoft is currently engaging neurodivergent employees via EY’s Neurodiverse Centre of Excellence in the testing of tools such as Copilot to ensure more inclusive design (e.g. determining where and how notifications should appear on the screen). Kellie Mote made the point that standards and compliance could also play a role: she suggested that a tool should only be considered “minimum viable product” if validated by a diverse user testing group, thus setting a benchmark for inclusion that products must clear before entering the market.

Encouraging the companies developing these tools to train developers about accessibility and bias. Previous OECD research (Touzet, 2023[46]) has called for computer science training to include modules on user experience design, human-computer interactions, and accessibility. An example of this highlighted in the report is Teach Access, a collaborative project between academia and the technology industry which aims to integrate accessibility principles into mainstream education for designers, engineers and researchers.

Use AI and other advanced technologies to help achieve more responsive, inclusive and innovative VET systems

AI and other advanced technologies can help VET systems react faster to changing skills needs, remove barriers to participation for diverse learners, and expand the repertoire of teaching, learning and assessment methods. Recent OECD work points to the value of data-rich guidance and modular learning offers, while country experiences show how digital platforms and skills-intelligence infrastructures can make programmes more dynamic and better connected to labour market demand. Insights from practitioners and employers further underline that, if thoughtfully deployed, AI can personalise pathways for neurodivergent learners and support smoother transitions into work. Governments can achieve more responsive, inclusive and innovative VET systems by:

Making VET more responsive with data and AI. VET systems can use AI, machine learning and big-data methods to detect changing skill needs faster, target updates to qualifications and steer provision to where demand is rising (OECD, 2025[54]; OECD, 2025[52]). Vocational rehabilitation expert Veronika Kaska stressed that training content must be updated regularly to reflect evolving tasks and tools. The OECD report on future‑ready VET (OECD, 2023[31]) argues that countries can enrich existing qualifications update processes with AI-supported analytics that identify not only new jobs but the specific competencies and regions where they are emerging, complementing but not replacing social-partner processes. Estonia’s reform of their qualification system illustrates this direction, by integrating forecasting with the qualifications system and developing digital solutions for skills registers and assessments. To turn this intelligence into action for learners, guidance services should use AI-enhanced labour-market information together with skills profiling and e‑portfolios, so learners can see which skills are in demand, map their current evidence to learning outcomes, and plan next steps as jobs and tools evolve (OECD, 2023[31]). Estonia’s national education information system, which links student records and outcomes and publishes aggregated data, shows how schools and learners can use evidence to navigate pathways.

Keeping VET relevant amid rapid technological change through employer engagement and work-based learning. VET stays relevant when learning takes place in authentic environments and when employers help to shape what is taught and how it is delivered. OECD work recommends engaging social partners to steer technology use, co-design sector toolkits and shared standards, and align AI-enabled content, assessments and equipment with workplace practice (OECD, 2023[31]). Rajesh Ananda warned that, as entry-level roles shrink, particularly in the tech sector, the real constraint is access to on-the‑job experience rather than the availability of tools. In the same vein, Esteban Tromel at the ILO argued for expanding and redesigning apprenticeships and other work-based learning so employers co-design curricula and assessments with providers, help select and trial tools, and co-deliver training and mentoring in real settings. This joint delivery makes it easier to adapt processes and technologies to individual needs, including those of neurodivergent learners, and ensures that what learners practise matches how work is actually done.

Using AI and advanced technology to broaden accessibility and inclusion in VET. In flexible VET systems, with modular pathways, micro-credentials, recognition of prior learning, and part-time or online study, AI and assistive tools can help more learners find and follow paths that are more efficient, better aligned with their competencies and prior experience, and closer to their goals and motivation. For instance, automated diagnostics based on e‑portfolios can help place learners into the right modules and recommend pathways. Day to day, built-in accessibility features powered by AI, such as real-time captioning, speech-to-text, text-to-speech, reading-level controls, plain-language rewrites and instant translation, make materials easier to access for multilingual learners and for neurodivergent learners who benefit from simpler language and clearer instructions. Executive‑function supports, including AI planners that break tasks into steps, visual schedules and smart reminders, help learners with ADHD manage practical work; conversational assistants can simplify rubrics and generate exemplars on demand without replacing human tutoring. Inclusion also depends on people and infrastructure: Ann Kennedy, a further education tutor, emphasised that inclusive use of AI should sit alongside targeted support, the development of staff expertise, smaller staff-to-learner ratios where needed and structured peer learning. Rohan Slaughter pointed to regionally funded assistive‑technology expertise to ensure consistent support in general education settings, and not only in SEN institutions.