How AI can be a panacea for the growing SEND, workload and retention crisis

REBECCA STOTT, DIRECTOR OF EDUCATION, TWINKL; NPQH; MA-ED, SEND & DIGITAL LEARNING, UK

GEORGINA DURRANT, SEND AUTHOR, BLOOMSBURY; NATIONAL THOUGHT LEADER FOR INCLUSIONAn approach where a school aims to ensure that all children are educated together, with support for those who require it to access the full curriculum and contribute to and participate in all aspects of school life, TWINKL, UK

JONATHAN PARK, AI LEAD, TWINKL, UK

Lost amid the noise of the election, the Department for EducationThe ministerial department responsible for children’s services and education in England updated its risk report in July to warn that the SEND (special educational needs and disabilities) system was ‘very likely’ to become financially unsustainable (DfEDepartment for Education - a ministerial department responsible for children’s services and education in England, 2024). Despite funding increases (IFS, 2024), with a rising number of pupils with SEND and OfstedThe Office for Standards in Education, Children’s Services and Skills – a non-ministerial department responsible for inspecting and regulating services that care for children and young people, and services providing education and skills pushing to close the attainment gap that widened during the pandemic (Stanley and Owston, 2024), educators are under mounting pressure – with 85 per cent reporting that ‘SEND pupils are not receiving adequate support’ (Teach First, 2024).

A recent poll, conducted by Twinkl SEND Digest (Brown, 2024), found that educators are facing a ticking SEND time bomb of growing demands and shrinking resources, contributing not only to the widening attainment gap but also to the workload and retention crisis – with 78 per cent of SENDCos (SEND coordinators) surveyed reporting that they have resigned or are considering doing so.

With limited fiscal levers to pull and long wait times for external services, government and education leaders need to prioritise EdTech, and AI (artificial intelligence) in particular, in their strategies to reduce workloads and improve outcomes.

Drawing on the research of the National Literacy Trust (2024), Education Policy Institute (2024), NFER/EFF (2024) and the DfE and National Audit Office (Davies, 2024), this article offers practical solutions on AI best practice, to support SEND learners, while reducing workload for educators.

AI and EdTech can provide useful tools for SEND practitioners and teachers as they face these growing challenges. However, as experienced SENDCo and author Georgina Durrant advocates, it is critical to remember that AI should never be seen as a replacement for human expertise. Like many technological advances before it, AI can be a powerful tool to support the teaching and learning of children with SEND, but it is vital that teachers use AI to assist, not replace, their expertise. Teachers’ professional judgement and deep understanding of individual learners’ unique needs are irreplaceable.

We need to ensure that we position teachers as the experts and leverage AI as a supplementary tool, enhancing teaching while ensuring that teachers’ experience and intuition remain at the forefront.

A whole-school approach 

Despite government backing of the use of AI in education (DfE, 2023), and with 75.3 per cent of educators (National Literacy Trust, 2024) recognising AI’s potential to reduce workloads and improve student outcomes, there are two key factors hindering the evolution of AI within schools: 

• A lack of training and support: three in four (75.3 per cent) teachers said that they also need more training, ICT support and resources in order to use generative AI tools effectively (National Literacy Trust, 2024) 
• Limited access to technology: Despite progress made during COVID, nearly half of schools (46 per cent) do not have reliable Wi-Fi access across their school, and a quarter of teachers do not have access to a school-owned device (tablet/laptop) that they can use at home for school work (Teacher Tapp, 2024).

Addressing the digital divide to improve outcomes, particularly those of the most disadvantaged (House of Lords, 2023), is not something that educators can tackle alone. The government, schools and trusts must remain committed to ensuring that all schools have access to high-speed internet (DfE, DDCMS and Zahawi, 2022) and relevant technology.

In terms of tackling educators’ skills and knowledge gaps, given the nature of technological evolution and rapid advancement, leaders should consider adopting a collaborative inquiry approach (Sharratt and Fullan, 2012) in order to empower educators within their settings. This will help to underpin all areas of school improvement with AI/EdTech best practice (DfE, 2023), without unnecessarily increasing workload – as opposed to approaching AI/EdTech as a standalone strategy or an add-on.

Applying the collaborative inquiry model

As with any whole-school change strategy, there are three key factors to prioritise for success: 

• motivation (Warner Burke and Litwin, 1992)
• adequate implementation and reflection time (Sharratt and Fullan, 2012) 
• the provision of ongoing impactful training, support and leadership (Leithwood and Riehl, 2003).

In terms of motivation, the first step to setting up a successful whole-school and/or trust collaborative inquiry model is establishing a meaningful inquiry question, linked to improving student outcomes, such as: 

How can AI be used in your area to improve outcomes for SEND learners?

It is then critical to allocate adequate time for research, discussion, relevant training, sharing of good practice (across the school or trust, in person or online), implementation, collaborative feedback and reflection (Sharratt and Fullan, 2012).

Supporting your team

As AI is a relatively new technology, creating a best practice landing page or online classroom is a great way of empowering educators by ensuring that they have instant access to the CPD (continuing professional development) that they need. An online hub can include short ‘how to’ videos that clearly demonstrate the use of the tools, relevant safeguarding information, links to interesting research, relevant longer-form CPD and a ‘wow wall’ for sharing AI wins.

Using AI today to improve outcomes for SEND and reduce educator workload

Since the introduction of the revamped curriculum in 2014, and the shift to a curriculum-focused Ofsted inspection framework in 2019, teachers, pupils and parents have reported increased workload (Churches and Fitzpatrick, 2023) and higher stress levels (BACP, 2023).

As E-ACT CEO Tom Campbell (2024) asserts, the impact of curriculum overcrowding goes beyond workload and wellbeing – it also impacts outcomes, particularly those of the most vulnerable and disadvantaged.

Expanding on this, Georgina Durrant describes how we are not meeting the needs of many learners with SEND through the current curriculum and assessment model. The curriculum is too narrow to allow all pupils to thrive and the academic pressures often act as a barrier to success. Children and young people deserve an inclusive education with a curriculum that allows for them all to succeed.

The solution, as Durrant and Twinkl AI Lead Jonathan Park assert, is adaptive teaching and personalised learning. However, as current workload statistics show (Churches and Fitzpatrick, 2023), finding the time to do this with over 30 children in every class, at a point when educators are already overstretched, is easier said than done.

Here, as research from the Education Endowment Foundation (EEF, 2024) and the National Literacy Trust (2024) illustrates, is where AI can help.

What tools are working? 

As outlined by the National Literacy Trust (2024), the percentage of educators who said that they had used AI increased from 31.0 per cent in 2023 to 47.7 per cent in 2024 in the following ways: 

• 56 per cent for ideas 
• 50 per cent to ask questions 
• 37.8 per cent to create lesson content 
• 34.7 per cent to generate model answers 
• 32.4 per cent for lesson planning 
• 26.7 per cent to make quizzes 
• 26.7 per cent to differentiate content
• 10.4 per cent to translate text 
• 4.7 per cent to assess work.

As the EEF summarises, this could support educators to reduce their lesson planning time by 31 per cent (EEF, 2024). 

Explaining in greater detail how AI can be used to support SEND learners, Jonathan Park suggests using AI for the following areas.

Adaptation of content for inclusion needs

Many teaching resources, particularly historic content, have not been designed with inclusion in mind. Moreover, the needs of children may often be highly individual, frustrating attempts to find resources that meet that need. Traditionally, teachers and educators have had to tailor this content themselves – a time-consuming process in a time-poor environment. AI and EdTech can play a key role in adapting content quickly, modifying resources to make them more accessible.

Personalised content to engage learners

Student engagement is ‘critical for successful learning’ (Wang and Degol, 2014, p. 137), and this is particularly true for some children with SEND. AI can be used to rewrite tasks to tailor them to the specific interests of the child, or to create reading comprehension activities focused on topics that excite the child, without teachers needing to spend time creating the resources themselves (National Literacy Trust, 2024).

Help and advice

Large language models (LLMs) have been trained on large amounts of text, including many discussions of SEND issues. As such, they can offer ideas and suggestions for how best to adapt content or a lesson for a child’s needs. However, caution is particularly needed in this area. AI tools can’t match the expertise of the classroom practitioner or the SEND expert, and they should only be used by trained professionals who can use the tools to spark thoughts on good adaptation.

What the future could look like

Predicting the future is always problematic, especially in the field of artificial intelligence, where technology can change quickly. However, looking at the trends of AI and machine learning, there are developments that could significantly shape the future of SEND teaching in the next 12 months, especially in the field of specific SEND-trained models. 

Most AI tools for SEND function either through prompt engineering on an LLM or, increasingly, through retrieval augment generation (RAG), where an LLM is required to reference an authoritative knowledge base beyond its training data sources, before it creates a response. Over the next 12 months, as greater steps are taken in fine-tuning, RAG and even purpose-built domain-specific models, the expertise and accuracy of the outputs of AI models will likely increase for SEND purposes.

However, as Jonathan Park asserts, AI is not an expert. LLMs respond based on the content on which they’ve been trained, which is not always drawn from knowledgeable sources. Think of it like a sounding board – good for ideas and getting a start but, ultimately, you’re the expert in your profession and your classroom.

The examples of AI use and specific tools in this article are for context only. They do not imply endorsement or recommendation of any particular tool or approach by the Department for Education or the Chartered College of Teaching and any views stated are those of the individual. Any use of AI also needs to be carefully planned, and what is appropriate in one setting may not be elsewhere. You should always follow the DfE’s Generative AI In Education policy position and product safety expectations in addition to aligning any AI use with the DfE’s latest Keeping Children Safe in Education guidance. You can also find teacher and leader toolkits on gov.uk .