Making AI work for students

New OfS and Advance HE research into the implications of AI for higher education brings together the perspectives of more than 300 students and academic staff. Josh Fleming, our Director of Strategy and Delivery, and Kathryn Harrison-Graves, deputy CEO at Advance CE, reflect on the findings.

Read the research report

This blog post was first published by HEPI on 11 November 2026.

“Ultimately, we've got to develop AI-literate graduates. You can't not do that.” (Staff research participant)
“And I think the jury's out at the moment about how we move... all students away from the kind of seductive approach that [AI] can do your work for you faster, to how you can use it as a real aid to learning.” (Staff research participant)

Artificial intelligence is transforming higher education. HEPI research suggests that 94 per cent of students use generative AI in the completion of assessed work. AI is now embedded in the search engines and academic tools students use every day, and graduates will compete for roles that increasingly require AI skills.

Institutions are innovating in response to these tectonic shifts. In our discussions with nearly 100 students and more than 200 staff, we heard about exciting AI applications that promise to personalise learning, widen access to support, create new opportunities for practice and free up staff time for meaningful interaction with students. But the pace of technological change, and the breadth of its implications, mean that responding effectively will be an ongoing challenge for institutions contending with a wide range of competing pressures.

Our research revealed a diverse and fast-moving picture. We found promising practice, alongside significant, shared challenges. Approaches also vary considerably, both between institutions and within them. By drawing on these experiences, together with published insights from across the sector, our work aims to give institutions a sector-wide picture of how others are responding as they navigate what comes next.

Key findings

Most students we spoke to use AI and many have integrated it into their studies in sophisticated ways. For some students, including some disabled students and students whose first language is not English, AI tools can support learning by breaking down barriers to access. Many of the students we heard from valued the frictionless way in which AI allowed them to ask questions and test ideas.

Similarly, many staff are innovating with AI in teaching, feedback, assessment, operations and administration, across subjects as varied as medicine, computing and photography. Many others are still in the early stages of exploring what AI tools can do.

Much of this echoes the existing literature. More revealing were the difficult choices AI presents for students and the complexities institutions face in deciding how best to respond.

We spoke to students who felt pressured to use AI, even when they weren’t confident it would support their learning, to help save time in busy schedules or because they saw it as the most reliable route to a good grade. Many worried about harmful ‘cognitive offloading’, the outsourcing of engagement with course materials or even assessments, and were frustrated that their institutions did not appear able to distinguish between human-authored and AI-generated work.

At the same time, most students we spoke to felt strongly that being AI literate would be essential to life beyond higher education. Looking to an uncertain jobs market, they wanted their institutions to help them develop the general and discipline-specific AI skills they believed employers would expect.

“Using LLMs has improved my learning in everything significantly, but at times it’s very easy to put multiple articles through it, just to get a summary, instead of reading full research papers. […] At times it can be very easy to sit back and let AI do the work for you.” (Student research participant)

With some exceptions, such as the launch of an AI Oversight Group and AI Hub at Anglia Ruskin University, innovation within institutions has primarily been driven by experimentation among individual staff members and departments. There are many pockets of innovative practice, but also big gaps in engagement with key questions at the institutional level, including on assessment. While many courses have adopted innovative approaches to ensure high-stakes assessments are robust, other assessments are vulnerable to completion using AI tools without students demonstrating the knowledge and skills being assessed.

What this means for the sector

Taking advantage of the opportunities and mitigating the risks AI presents will require a strategic approach. It will depend on senior leaders considering the implications of AI carefully and working with students and staff to drive change.

Assessment will require urgent and ongoing attention. Effective assessment will often require forms of assessment designed to prevent students from using AI, such as exams, presentations, performances or practical assessments. Assessments where AI is permitted will need careful review to ensure they effectively test the intended learning outcomes. The University of Bath, for example, has improved the resilience of its assessments through a ‘two lane model’ which requires assessments to either take place in a closed, AI-free environment or be designed assuming students will use AI. Irrespective of form, assessment should be accessible and inclusive as well as valid and reliable, so that every student has a fair chance to demonstrate their knowledge and skills.

Our next steps

Institutions and academics are best placed to understand students' changing needs and determine what an effective response to AI looks like in their own contexts. We are also mindful that heavy-handed intervention could constrain innovation in a rapidly evolving environment.

We don’t have all the answers and the technology will continue to develop. But the speed and scale of the changes brought by AI means that waiting for a settled picture is not an option. The OfS will continue working collaboratively with institutions and sector bodies to support an effective response while facilitating an environment conducive to innovation. This includes:

In response to the research findings, the OfS will:

  • Continue to deepen sector understanding of how AI is reshaping higher education by undertaking further structured research to establish the scale and nature of change.
  • Support universities and colleges to strengthen their approaches to AI, including by clarifying how OfS regulatory requirements apply in the context of AI and sharing examples of effective practice.
  • Take action through the OfS’s existing regulatory tools where we identify risks to students’ interests. We will not introduce new regulations.
  • Under the proposed future approach to quality, we would expect institutions to demonstrate in their provider submissions that their assessments remain valid measures of students’ knowledge and skills in the context of AI.

To help institutions develop their approaches to AI, Advance HE will:

  • Launch an AI focused Change Academy programme for members institutions early in 2027. Designed as a peer-learning and action-learning programme connecting Advance HE members from across the globe, the Change Academy will create space for leaders to pause, reflect and develop their institutional AI strategy.
  • Convene sector leaders, industry experts and academics to explore effective approaches to responding to AI within teaching and learning, leadership, governance and inclusion.
  • Continually review frameworks and resources available to members via the Advance HE Knowledge Hub to provide guidance and shape discussion around ethical and appropriate use of AI in higher education.  

As we move into the next phase of our work, we want to hear from you. On Monday 16 November, the OfS and Advance HE will hold a joint webinar discussing the findings and considering how institutions can take our findings forward. We hope you’ll work with us in the coming response to build an innovative, rigorous approach to AI in higher education.

Read the research report Find out more about our work on AI in higher education

Comments

There are no comments available yet. Be the first to leave a comment.
Leave a comment
*
*
*
Published 18 November 2026

Describe your experience of using this website

Improve experience feedback
* *

Thank you for your feedback