The Great Equaliser: Can AI Close the Inclusion Gap in Education?
AI in Education Part 1 | Reading time: approx. 8 minutes.
There is a question sitting at the heart of almost every staffroom conversation about artificial intelligence right now, and it is not “should we be using it?” That ship has sailed.
The real question is far more important: who will benefit, and who will be left behind?
The optimistic answer, and the one increasingly being put forward by policymakers, EdTech companies and educators, is that AI has genuine potential to close the inclusion gap. To give every child, regardless of their postcode, diagnosis or background, access to personalised, responsive learning support that was previously available only to those who could afford it.
But is that promise realistic?
This is the first in a three-part series exploring that question honestly. Here, we look at the promise of AI for inclusion. In Part 2, we will consider the risks and the potential for AI to deepen existing inequalities. In Part 3, we will look at what a genuinely equitable future for AI in education might actually need to look like.
What the Policy Framework Is Telling Us
The Department for Education (DfE) has been clear about its ambition. Its Generative AI in Education guidance, first published in 2023 and updated in January 2025, identifies the potential for AI to reduce workload for teachers and personalise learning for pupils, including those with SEND (Special Educational Needs and Disabilities).
The guidance acknowledges that AI tools “could offer significant benefits... including helping to identify and meet the needs of pupils with SEND” (DfE, 2025).
The government's Roadmap for Modern Digital Government: AI in Education goes further, setting out a vision in which AI-powered tools can improve pupil outcomes while reducing the administrative burden on educators. The principle is simple: if technology can take some of the repetitive work away from teachers, that time can potentially be redirected towards the children who need it most.
Perhaps most significantly, in 2025 the government announced that up to 450,000 disadvantaged pupils could benefit from AI tutoring tools as part of a wider initiative to use technology to help close the attainment gap.
This is not a small-scale experiment.
It signals a clear institutional belief that AI could form part of the answer to one of the most persistent challenges in English education.
A Tool That Meets the Child Where They Are
For those of us working with pupils with SEND, perhaps the most compelling aspect of AI is not the technology itself. It is the responsiveness.
The SEND Code of Practice (DfE/DoH, 2015) places a statutory duty on schools to identify and address barriers to learning, provide “high quality teaching that is adapted and differentiated” (6.37), and ensure that the graduated approach of Assess, Plan, Do, Review underpins intervention.
At its best, AI has the potential to help make this approach more responsive and more achievable at scale.
Consider what it can offer:
Personalised learning pathways. Adaptive learning platforms can adjust the pace, format and complexity of tasks in real time, responding to individual performance in a way that is difficult for a single teacher to replicate across a class of 30. For pupils with dyslexia, dyscalculia, ADHD or autism, dynamic scaffolding can potentially remove barriers without making the learner feel different from their peers.
Assistive and augmentative technologies. AI-powered text-to-speech, speech-to-text, predictive text and real-time captioning tools are already changing what is possible for pupils with physical, sensory and communication differences. Increasingly, these tools are built into mainstream devices, making them easier to access and, importantly, less likely to mark a child out as needing something different.
Communication support. For non-verbal pupils or those using AAC (Augmentative and Alternative Communication), AI is beginning to support more natural and contextually appropriate communication. Tools that can learn a child's vocabulary preferences or communication style represent a genuinely interesting new development.
Supporting EHCPs and review processes. Schools Week reported in 2024 that DfE guidance now suggests schools might use AI to assist in drafting SEND support plans. Used appropriately, and with careful human oversight, this could free up valuable SENCo time that might otherwise be spent on paperwork and allow more time for direct work with pupils, families and staff.
There is also a wider legal context here.
The Equality Act 2010 requires schools to make reasonable adjustments to avoid putting disabled pupils at a substantial disadvantage. As AI becomes increasingly embedded in the tools pupils use every day, it is likely that AI-supported technology will increasingly form part of the conversation around what constitutes a reasonable adjustment.
The Tutoring Gap and AI as a Leveller
One of the clearest inequalities in English education is access to private tuition.
Research has consistently shown that pupils from higher-income families are more likely to receive one-to-one or small-group tutoring. This additional support can supplement what happens in school and, over time, compound existing advantage.
AI tutoring platforms offer, at least in theory, a way of democratising access to this kind of support.
When designed well, an AI tutor can provide patient, adaptive and low-stakes feedback without the cost associated with a human tutor. It can allow a pupil to ask the same question several times, try something again, work at a different pace and receive immediate feedback without the fear of getting something wrong in front of their peers.
That matters.
The government's National Tutoring Programme (NTP), while not AI-specific, established the policy logic that targeted tutoring can support disadvantaged pupils. AI could potentially provide a way of scaling that principle much further.
Third Space Learning and similar platforms have demonstrated that remote, AI-supported tutoring can produce measurable gains in maths attainment for disadvantaged primary school pupils. These findings are contributing to a growing evidence base around the potential of technology-supported tutoring.
But this is still an emerging evidence base.
The Nuffield Foundation has invested in research into AI and education precisely because it recognises that there is significant potential for both opportunity and harm. It has called for rigorous, independent evaluation of AI tools before widespread adoption. That feels like an important position to take. The promise is significant, but so is the need to understand what actually works, for whom, and under what conditions.
What the Evidence Is Starting to Show
A 2025 systematic literature review published in Discover Artificial Intelligence (Springer Nature) found that AI-based personalised learning interventions showed positive effects on student engagement and learning outcomes across multiple studies. The authors also highlighted an important limitation: evidence relating specifically to high-need and SEND populations remains limited, and effect sizes varied considerably depending on the context and the quality of the tool.
That distinction matters.
It is easy to talk about AI as though the technology itself is the intervention. It isn't. The impact depends on how the technology is designed, how it is introduced, what data it uses, who has access to it and how effectively it is integrated into teaching.
Written evidence submitted to the House of Commons Education Committee's inquiry into AI in Education (2024-25) from a range of organisations, including disability organisations and research bodies, similarly recognised both the opportunity and the risk.
The Digital Futures for Children Centre noted that AI tools could be “genuinely empowering for pupils with disabilities”, but only when those pupils have a meaningful role in their development and when robust safeguarding frameworks are in place.
In other words, inclusion cannot simply be something that happens to children through technology. Children need to be part of the conversation about what is being developed and how it is used.
The Human-Technology Partnership
There is something important that can get lost in the more enthusiastic conversations about AI.
AI does not replace the relational heart of inclusion.
The SEND Code of Practice 2015 is built around the principle that families, young people and professionals work together in partnership. Ofsted's new inspection framework (2025) continues to emphasise the quality of teaching, relationships and pastoral care as the foundation of effective schooling.
AI is a tool.
Its value depends on the humans who use it, the intentions behind its use and the structures within which it sits.
A well-designed AI writing support tool, used within a culture of high aspiration and genuine inclusion, could be transformative.
The same tool, introduced into an under-resourced school without appropriate training, oversight or understanding, could simply produce superficially compliant provision that looks efficient while leaving underlying needs unmet.
That is the distinction we need to keep coming back to.
The promise of AI for inclusion is real.
It could give more children access to personalised support. It could reduce barriers that have historically been difficult to overcome. It could give teachers back time. It could make some forms of tutoring and assistive technology more accessible.
But none of that will happen automatically.
And it certainly will not happen equally without deliberate effort, appropriate safeguards, staff training, policy support and investment.
The question, therefore, is not simply whether AI can make education more inclusive.
It is whether we choose to use it in a way that actually does.
Part 2 will look at the other side of the equation: the peril of AI for inclusion, and the very real possibility that a technology designed to close gaps could, without careful thought, make some of them wider.
References and Further Reading
Department for Education (2025). Generative Artificial Intelligence in Education. Gov.UK. Read here
Department for Education / Department of Health (2015). SEND Code of Practice: 0 to 25 Years. Gov.UK.
Department for Education (2025). AI in Education — Roadmap for Modern Digital Government. Read here
Equality Act 2010. Chapter 15. legislation.gov.uk.
Equality and Human Rights Commission (2015). Technical Guidance: Reasonable Adjustments for Disabled Pupils. Read here
Gov.UK (2025). 450,000 Disadvantaged Pupils Could Benefit from AI Tutoring Tools. Read here
Nuffield Foundation. Artificial Intelligence and Education [Research project]. Read here
Schools Week (2024). DfE Guidance Suggests Schools Use AI to Draft SEND Support Plans. Read here
Digital Futures for Children Centre (2024-25). Written evidence to the House of Commons Education Committee AI in Education Inquiry (AIE0181). Read here
Springer Nature / Discover Artificial Intelligence (2025). AI-based personalised learning in education: a systematic literature review. Read
