Introduction
Over two roundtable events, London in May, Manchester in June, we brought together colleagues from across further and higher education to talk about how AI is shaping accessibility and inclusion. From the tools students are using, to the policies (or lack of) that guide them, and the kind of future we’re all trying to create, the conversations were open, insightful and, at times, challenging.
The first focused on the relationship between AI and inclusion, exploring whether AI might not only reinforce existing inequalities but also offer opportunities to address them. Drawing on experience working in schools in areas of high deprivation, Tom Moule (Senior AI Specialist) highlighted the persistent barriers some students face in accessing devices, data, and supportive learning environments. He asked whether, beyond equalising access to AI, we might also see it as a tool to make learning more inclusive. Examples included using AI to help write cover letters or personal statements, forms of support that often rely on social capital or paid help. As colleges and universities rethink assessment in response to generative AI, the provocation was clear, will we use this moment to take a more inclusive view of students’ skills and how we recognise their value?
The second, delivered by Helen Nicholson-Benn (AI Specialist and Subject Specialist -Assistive Technology), considered the relationship between AI and accessibility. She began with acknowledging how AI has long played a critical role in many assistive technologies, from screen readers to digital assistants, highlighting that disabled users have often been at the forefront of adopting AI technology in daily life. The presentation showcased a range of innovative uses of AI to support accessibility, both from purpose built assistive tools such as AI note takers, and of ‘general purpose’ chatbot tools which users are increasingly exploring independently, outside of formal support systems. This increasing potential was contrasted with some rising concerns including access issues to AI tools themselves, premium support features being hidden behind paywalls, and the added challenges disabled learners face due to unclear or restrictive institutional policies. This raised the question to attendees of how to effectively build on the support AI can offer, while mitigating risks and making learning environments more inclusive by design.
This report captures the key themes that emerged across both events: what’s working, what’s not, and where we need to go next. While there was plenty of common ground between the two cities, each brought a distinctive focus with London leaning into systemic questions and philosophical implications and Manchester more concerned with practicalities and implementation.
The conversations across both roundtables were rich, often surfacing insights we can’t afford to ignore. As we continue to shape our work, we now need to reflect on what we’ve heard and consider how these perspectives can be meaningfully embedded in what we do next.
Below are the key insights and takeaways that emerged from these conversations.
- Tools: Benefits, risks and realities
There was broad agreement that AI tools are already playing a significant role in education. Translation tools and grammar checkers, such as Grammarly and Google Translate, came up repeatedly. These are proving particularly helpful for international students and others who struggle with academic English. For many, they act as an invisible support, helping students to understand, complete, and improve their work in ways they might not otherwise manage.
But alongside the benefits came real concerns. Some students worry they’re becoming too reliant on AI and aren’t developing their English skills as fully as they should, as they rely on AI to do the heavy lifting. For others, the fear is about being accused of cheating, flagged as using AI inappropriately, particularly by detection tools that don’t understand the nuance of neurodiverse and second language writing styles.
Staff are grappling with similar tensions. Many remain unsure where the line falls between acceptable support and inappropriate use, especially in creative subjects, where the idea of ‘authentic voice’ matters. One example from a textiles course stood out: a student felt that using generative AI stripped the identity from their work. The tool produced the image they wanted, but it no longer felt like theirs. That tension, between assistance and replacement, echoed in different ways, across both roundtables.
There’s also a growing divide in digital literacy. The most confident students are using AI to their advantage, refining prompts, speeding up tasks, and tailoring outputs. Others are muddling through and using tools in ways that weaken their learning. While students have the right to not engage, it’s important to recognise that AI is likely to play a role in their future careers, and opting out entirely may limit their opportunities. The challenge is not just giving access but building confidence and capability, so students can make informed, empowered choices. Everyone agreed: students need proper training on responsible, context-specific use, not just access.
Access itself is uneven. Institutions reported inconsistent availability of tools like Gemini, Copilot, and Adobe Creative Cloud. Some students have access, others don’t, depending on course, licensing, or institutional budget. In creative subjects, while staff attitudes are shifting, uncertainty still lingers about what’s acceptable. This inconsistency is creating confusion and insecurity for both students and staff. Many educators haven’t had formal training and are learning through trial and error. Meanwhile, students and learners are asking: ‘Is using Grammarly fair? Should we be taught how to use it properly?’ These questions point to a deeper issue, not just about access to tools, but about a shared understanding of how AI should be taught, supported, and assessed.
Cost and equity were raised particularly strongly in Manchester. Many AI tools require paid versions to unlock essential features, leaving some students and staff behind. Even where AI has been embedded. like via Canvas integrations, students and learners often don’t trust these institutional tools, unsure who can see their data or how it might be used.
In London, discussion focused more on ethical concerns. Should students and learners be expected to use AI tools they fundamentally object to? Some raised discomfort with platforms linked to major tech firms like OpenAI, citing issues around data privacy, sustainability, and corporate influence. As AI is embedded more deeply into everyday platforms like Microsoft Word and VLEs it becomes harder to opt out. Yet meaningful alternatives are rarely offered.
These concerns raised wider questions about how AI tools are chosen in the first place. Is enough thought being given to whether tools are designed with all learners in mind? Or are they optimised for those students and learners with strong digital skills, cultural familiarity, and fewer accessibility needs?
- Processes: Gaps, Tensions and What’s Missing
When it came to policies and processes, both roundtables painted a similar picture. Most institutions are reactive rather than strategic, focusing heavily on assessment and plagiarism, but saying little about accessibility or inclusion. There’s no real consistency, either between or across institutions. Some departments are pushing ahead, others are doing very little, and students are often left confused.
One of the biggest concerns was the lack of clear, usable guidance. Students said policies are hard to find and even harder to understand. They want something simple and direct, but instead they’re often faced with vague, technical documents written in formal or academic language. This can present a specific barrier for neurodivergent students and those who speak English as an additional language, who may find dense or abstract language particularly difficult to navigate. Staff don’t always feel much better. Some are using AI tools but are afraid to admit it, others avoid them altogether because they don’t feel confident or supported.
The discussions surfaced discomfort around who shapes policy. Too often, it’s one or two people, without involving those most affected. There might be student and learner involvement in writing guidance or running workshops, but engagement in actual decision-making is rare. Both events made clear that this needs to change, and not just symbolically. If disabled, neurodivergent, multilingual and international students are left out of policy design, then those policies will never fully work for them.
A strong theme in London was the role of language in shaping inclusion and exclusion. Participants questioned whether universities should introduce explicit language policies that value multilingualism, rather than defaulting to narrow standards of academic English. Institutions often claim to value diversity, but this rarely extends to how language is handled. Expectations around correct English can disadvantage students who think or write in other languages or dialects. With AI now embedded in many tools, and government backing wider use across education, it’s time for colleges and universities to adapt.
Another challenge is the language used around AI. Terms like prompt engineering can feel alienating, especially for students and learners with lower digital confidence or those for whom English isn’t a first language. As one participant put it, you can just ask AI to write the prompt for you,but that knowledge isn’t being shared. Making AI more accessible means rethinking the language we use to describe it, not just the tools themselves.
Manchester, meanwhile, was more focused on the nuts and bolts of how to make policies clearer, how to get the right people in the room, and how to train staff who feel behind the curve. There were some positive examples too: institutions trialling AI in assessments with clear boundaries, student-led hackathons, and practical efforts to reshape assessment formats.
- The Future: Possibilities, Pressures and Priorities
Looking ahead, both roundtables made it clear, if the sector doesn’t act now, AI will widen the very gaps we’re trying to close. Without thoughtful planning, disabled students, international students, and those with lower digital confidence or limited access to tools risk being left behind. Some are already feeling it.
AI isn’t going away, it’s already embedded in student life, in workplace expectations, in how knowledge is accessed and expressed. Employers increasingly value AI skills over formal qualifications, and students are picking up on this. If institutions don’t build AI capability into learning, they risk graduating students without the skills they’ll need to thrive.
There are also big implications for the UK’s international student pipeline. Many students come here to earn a prestigious degree, to master academic English, and to improve their communication in a global language. But now, with AI able to instantly translate, edit, and polish English to an academic standard, some are questioning whether a UK-based education is still the unique offer it once was. If we don’t adapt, the sector’s dependence on international recruitment could become a vulnerability.
There’s also a risk that AI will widen the digital divide, unless steps are taken to ensure all students and learners, not just the most confident, can use it well. Currently, AI is helping some students and learners gain an advantage, while others fall further behind. Yet AI holds real potential to support accessibility and inclusion across both colleges and universities. It can help structure thoughts, simplify complex information, and act as a consistent, non-judgemental support. For many learners, it’s already filling gaps in systems that haven’t always worked for them.
But for that potential to be realised, developers need to understand accessibility, not just as a compliance requirement, but as a core design principle. Tools must be built with the full range of users in mind, not just those who already thrive in academic settings. And institutions need to take a more student-centred, consistent approach, ensuring that policy, support, and access aren’t left to chance or departmental preference.
Language remains a critical theme. Continuing to penalise non-standard English or multilingual expression only reinforces inequality. As one participant noted, being able to express ideas clearly in standardised written English is a form of privilege. Not everyone has equal access to the linguistic norms expected in academic and professional spaces, particularly international students, regional English speakers, and neurodivergent learners. For these groups, generative AI is not a shortcut, it’s an essential tool for linguistic accessibility. Participants also called for an end to AI detection tools, which unfairly penalise students who write differently. It’s time to rethink how we define originality and authorship, and to do so in ways that are genuinely inclusive
More broadly, policy processes need to become far more dynamic. Technology is moving fast, and the previous slow-to-update policies simply won’t keep pace. Instead, we need flexible frameworks, co-designed with students, that focus on inclusion, and adaptability. That includes recognising new forms of learning and assessment and embracing a wider range of ways to demonstrate knowledge and understanding.
The technology itself will keep evolving. We’re already seeing wearable, AI-powered assistive technologies on the horizon. These bring incredible opportunities, but also new risks around privacy, data use, and inequality of access. Institutions will need to stay ahead, not just with tools, but with thinking.
Both groups agreed that whatever comes next, it must be inclusive by design. We can’t afford to bolt on accessibility or inclusion later or assume one size fits all. The future of AI in education will be shaped by the decisions we make now, not just about tools and policies, but about values. Are we building a system where all students can thrive, or one that leaves the same groups behind?
Closing Thoughts
The future of AI in education isn’t just about innovation, it’s about values. If we want an inclusive, future-ready sector, we need to act with intention, clarity, and urgency. That means investing in skills, demanding inclusive design from developers, and embedding equity into every decision, from classroom tools to national policy.
These roundtables showed a sector in flux, still experimenting, still learning, but also clearly committed to getting this right. Students want support, not suspicion. Staff want clarity, not confusion. And everyone wants the freedom to use AI in ways that help, not hinder.
The biggest takeaway? AI isn’t the issue, it’s how we choose to implement it that will decide whether it becomes a driver of inclusion or yet another barrier to overcome.
Thanks to participants:
Chichester College Group, Rebecca McCardle
De Montfort University, Sultan Chaudhury
Gloucestershire College, Freya Bevan
Goldsmiths, University of London, Jennifer George
Natspec, Fil McIntyre
Ulster University, Brian McGowan
University of Bristol, Sarah Davies
University of Oxford, Kelly Webb-Davies
University of Sheffield, Laurie Wilson
University of the Arts London, Xavier Briche
University of the Arts London, Kei Ferguson
World Skills, Rebena Sanghera
York College & University Centre, George Pickard
Find out more by visiting our Artificial Intelligence page to view publications and resources, join us for events and discover what AI has to offer through our range of interactive online demos.
Join our AI in Education communities to stay up to date and engage with other members.
Get in touch with the team directly at AI@jisc.ac.uk