Written evidence by Dr Frankie Rogan and Dr Kamilya Suleymenova (AIE0200)

Education Committee

The use of Artificial Intelligence and EdTech in Education inquiry

Introduction

We submit this evidence to the Education Committee as two academics (Dr Kamilya Suleymenova and Dr Frankie Rogan) working in education leadership roles at the University of Birmingham (UoB). Below, we briefly introduce ourselves and our experience in working on issues of AI within higher education to provide some context for why we are well placed to submit this evidence to you.

Since April 2025, Dr Kamilya Suleymenova has acted as Deputy Director of Education for the College of Social Sciences at UoB, with a specific remit for Digital Education. Kamilya has extensive experience in dealing with issues of AI in higher education and was previously ‘GenAI Lead’ for the Birmingham Business School. This role provided dedicated time and space to thinking about the role of AI (specifically generative AI) in pedagogy and practice. In this role, Kamilya developed a programme of training and curated resources for colleagues to support them in their understanding of the rapidly developing terrain of GenAI and also implemented basic GenAI awareness for all students within the Business School. In her current role as Deputy Director of Education for the College of Social Sciences, Kamilya is working to extend this work by supporting colleagues with AI literacy across the other three Schools within the College (School of Education; School of Government; School of Social Policy and Society).

Dr Frankie Rogan has been the Head of Education for the School of Social Policy and Society at the University of Birmingham since August 2024. In this role, Frankie has oversight of all education matters within the School and this has inevitably meant navigating the questions and concerns that both staff and students have about the role of AI in higher education and, more specifically, social science disciplines. She was also Deputy Head of Department (Education) for the Department of Social Policy, Sociology and Criminology between 2021 and 2024, overseeing education provision at departmental level. In her roles, she has observed the rapidly changing environment of AI and education and has supported staff to think differently about assessment and curriculum design within the new ‘Age of AI’. At a national level, she has recently been on the academic advisory group for the revised QAA Subject Benchmark Statement for her discipline, Sociology. The revised statement included the development of a section on Generative AI for the first time.

Along with colleagues, we attend and represent staff and student interests at a range of meetings and committees at School, College and University level. These include broader Education and Quality Assurance Committees, as well as more specifically focused committees and/or working groups, such as a ‘Special Interest Group for AI in Education’. As such, we are well placed to submit evidence to this committee, having extensive experience in navigating the challenges, opportunities and tensions that emerge in discussions relating to AI in higher education.

The below statement is structured around five key areas identified in the call for evidence: staff and institutional knowledge and training; the digital divide; assessment; skills and curriculum; and implementation.

Staff and institutional knowledge and training

Higher Education is often structured around disciplinary specialisations and developing training/skills that cut seamlessly across disciplines (such as effective and critical use of GenAI) does not sit naturally within this structure. With the rise of AI, staff are being asked to rapidly develop competencies that fall outside their disciplinary identity and don’t fall into traditional understandings of an academic job role. For many academics, engaging with AI raises questions about what their expertise means, how their role is defined and how they effectively develop students’ critical thinking skills in a rapidly developing environment increasingly shaped by student engagement with AI. This makes ‘upskilling’ of staff not just a challenge of training and time but also a broader challenge of identity as lecturers navigate the new terrains within which students learn. High workloads in academia also make it difficult for staff to find adequate time to adapt appropriately to these shifting terrains and support for developing staff knowledge in this area is often spread unequally across disciplines. Staff responses to issues of AI in teaching and learning are also highly divided and therefore rolling out a universal type of ‘training’ and/or ‘upskilling’ is challenging without clear guidance. Staff have a range of views and attitudes on AI, with some raising understandable concerns about the biases embedded into specific forms of AI, impacts on sustainability and impacts on academic integrity, ethics and ownership.

It is also important to note that higher education institutions often operate within complex quality assurance frameworks, regulatory requirements, and expectations around student support, pastoral care and safeguarding. These processes exist for good reason, but they do make change and adaption to fast-moving contexts difficult, as changes to assessment practices and ways of teaching and learning need to consider a range competing concerns. Quick and uncritical adoption of new technologies (such as AI) is inherently contentious and universities are expected – by students, regulators and the public – not to take risks with their educational provision. This can make adapting and responding to developments in AI increasingly tricky. Supporting the creation of regulated ‘test-zones’ within universities – spaces where regulatory requirements and the ordinary processes to change assessment and learning outcomes are relaxed to enable controlled experimentation with teaching and assessment would allow institutions to learn what works before attempting to scale-up too rapidly in ways that aren’t feasible or appropriate.

The digital divide: access and skills

Students arrive at university from vastly different backgrounds and have (and will continue to have in the future) very different starting points in relation to their understandings, use and views of AI. This can be shaped by several factors, including disparities in experiences at school and home before they arrive at university. School equipment and infrastructure vary considerably in what technology is available to students (and at what stage), and school policies on AI are also inconsistent. This means that some students arrive at university already regularly using AI, while others have had far less exposure. While universities can attempt to address this through making training and resources available and introducing students to critical reflections on AI usage within generic academic skills modules, the gap is already significant by the time students arrive and, if curricula and assessment do not adapt adequately, this divide may risk exacerbating awarding gaps and graduate outcomes between the most and least privileged students.

A baseline requirement for digital literacy, including critical engagement with GenAI (e.g. understanding what it can and – crucially – what it cannot do and what ethical considerations need to be considered during its use), at school level may reduce the disparity in preparedness among students entering higher education. Guidance should be clear and nationally consistent to develop a clearer baseline of knowledge and experience before entering university.

Assessment

Assessment in the age of AI has become a central site of tension within universities, where staff concerns and anxieties about AI, pedagogy, and academic standards are most often expressed. Academic staff often hold divergent positions on the role of AI, ranging from those who wish to forbid its use altogether to those who seek to integrate it fully into teaching, assessment design, and learning outcomes. For students, this lack of consensus frequently results in inconsistent or unclear guidance, with institutional policies and module-level expectations sometimes appearing vague or even contradictory in defining what constitutes appropriate use of AI in academic work.

Within many disciplines (particularly those traditionally reliant on essay-based coursework), AI is often framed as a threat to established forms of written assessment. This has prompted calls in some areas for a return to timed, unseen examinations as a more secure and controllable method of evaluating student performance. However, such a shift raises pedagogical concerns. The limitations of relying solely on exam-based assessment have been widely recognised within higher education over a number of years. In recent years, considerable effort has gone into varying assessment methods to better reflect a wider range of skills, to enable students to communicate with a wider range of audiences and to prepare them more adequately for a range of graduate careers. A wholesale return to unseen examinations risks undermining this progress, while also re-entrenching well-documented issues such as awarding gaps and inequities in student outcomes.

These debates are shaped by the practical realities of AI use, which are often understood in binary terms. Despite guidelines given to students across different universities and departments (e.g. ‘traffic light’ systems that signal how much – if any – AI usage is permitted within any given assessment), in practice, assessments tend to fall into two broad categories: those conducted under controlled conditions, such as invigilated exams, oral assessments, or live presentations, where the use of generative AI can be restricted; and those completed outside of a controlled environment, where AI use is effectively impossible to police and must therefore be acknowledged and incorporated into assessment design and marking criteria. We must also ensure students are able to be transparent about their use of AI – if it has been used, how it has been used and why it has been used. This encourages students to avoid uncritical and unreflective use of digital technologies.

Assessment in higher education has traditionally served two key functions: as a process of learning and as a mechanism of certification. As a form of learning, assessment is intended to develop students’ understanding and knowledge, communication skills, and capacity for critical thought through the act of completing the assigned task. The availability of generative AI potentially alters this process, raising questions about the extent and nature of the cognitive work undertaken by a student when a piece of work is submitted. As a form of certification, assessment signals to employers and to students themselves what an individual knows and is able to do. If AI significantly supports the production of assessed work, the reliability of this signal may be weakened. These two dimensions of assessment are disrupted in different ways by AI and therefore require distinct, carefully considered responses.

A further challenge lies in the pervasive uncertainty surrounding the future role of AI in work and society. Universities cannot yet determine with confidence which AI-related skills will be most valuable, for whom, or in what contexts. As a result, it is difficult to define what should be assessed or what kinds of skills assessment should signal. This is not a problem that can be resolved through effective assessment design alone, as the technological landscape is evolving too rapidly. Instead, it calls for ongoing dialogue between higher education institutions, policymakers, employers and wider society to ensure that assessment practices remain relevant and meaningful.

In response to this uncertainty, many institutions have adopted a cautious strategy of assessment diversification, seeking to reduce reliance on any single method by employing a broader range of formats. While this approach has merit, it also introduces trade-offs, including increased workload for both staff and students, potential incoherence across programmes and less opportunity to ‘master’ a particular skill through repetition (e.g. essay writing). Moreover, as AI technologies continue to advance, it is becoming increasingly difficult, particularly in the humanities and social sciences, to identify forms of assessment that cannot be meaningfully assisted by AI. Again, this necessitates a holistic rethink of assessment and curricula for a new age, rather than knee-jerk quick fixes.

There is also a critical equity dimension to consider. For some disabled students, AI tools can function as powerful assistive technologies, offering forms of support that go beyond traditional accommodations and potentially enabling more equitable participation in assessment. Policies that impose blanket restrictions on AI may inadvertently exacerbate these inequalities. Any approach to AI in assessment must therefore be attentive to issues of accessibility and inclusion.

Finally, it is important to recognise that the impact of AI on assessment is not uniform across disciplines. The challenges and opportunities it presents differ significantly between fields such as the social sciences, humanities, and STEM subjects. Effective policy and guidance must therefore allow space for disciplinary expertise and judgement, rather than imposing overly rigid, one-size-fits-all solutions.

Skills and curriculum

AI represents a kind of ‘revolution’ in information processing. Just as the industrial revolution required a fundamental shift in the skills of the workforce, GenAI may demand an equivalent shift, but this time centred on how we process, evaluate, and create with the information available to us. The intensity of information and productivity expectations have risen to a point where managing them requires distinct, deliberately developed skills: human-centric interpersonal skills, the ability to ask good questions, and the capacity to evaluate information quickly and critically. Most previous education and training was structured around the ability to perform tasks that would be repeated and improved over time: solving similar problem sets, writing essays to develop writing, citation and communication abilities, and so on. AI fundamentally changes this– anything replicable can potentially be handled by AI. What becomes more valuable is creativity, originality, and the ability to think at a systems level.

As such, systems thinking should be taught more explicitly. Currently, university courses are usually structured around single disciplines or joint honours combinations. Teaching is delivered per subject, per module – effectively training students to compartmentalise knowledge rather than connect it. Proper ‘deep’ interdisciplinarity is key to developing the capacities that GenAI makes most valuable, yet current structures can actively work against enabling it.

The Curriculum and Assessment Review should be examined for whether it adequately anticipated the pace and depth of AI adoption. There is likely a need for further focused work in this area, particularly on how curricula can be designed to foster the integrative, creative, and critical skills that AI now makes essential.

Implementation

Finally, there are also challenges in terms of implementation of AI guidance in relation to education within universities. The most challenging of these operates at the meso-level – between broad institutional strategy and guidelines and local interpretation of such guidance within specific modules and assessment designs. It is relatively straightforward to write institutional policies once a particular direction is agreed. It is also manageable for individuals with the relevant skills to adapt specific modules or assessments in line with that guidance. But translating strategy into operational practice across an entire School or department – with limited resources and competing priorities – is where implementation most often stalls.

This is where networking and practice-sharing should be concentrated: at a level that is more than individual case studies but less than institutional strategy. Cross-institutional communities of practice at this intermediate level would be valuable. Greater engagement with employers and wider society is also needed to inform what skills and competencies are genuinely required, and to ensure that any changes to provision are appropriate and proportionate. Universities also have a civic duty that extends beyond degree-level education: to help people understand how to treat information and evidence critically and to continue to reflect on and develop the skills acquired through higher education throughout their life. The assumption that someone could complete a degree and then perform broadly the same role for most of their career is probably no longer valid. Universities should be supported to play a role in lifelong learning and public information literacy.

Overview

This evidence draws on sustained engagement with large cohorts of academic staff and students, alongside formal participation in university committees responsible for education, assessment, quality assurance and policy development. It therefore reflects both local and institutional insight into how AI and educational technologies are reshaping higher education in practice. Across these domains, a consistent picture emerges of both significant opportunity and profound challenge. AI provides an opportunity for universities to rethink their programmes, including what skills we are developing and assessing, as well as how to better enable student creativity and originality within the context of AI. However, it also exposes structural tensions within higher education systems, particularly in relation to staff capacity and training, uneven student knowledge, the integrity and purpose of assessment, and the coherence of curriculum design. These challenges are compounded by the pace of technological change, which outstrips the caution that is often required by existing regulatory and pedagogical frameworks.

We emphasise that effective responses cannot rely on isolated or short-term solutions. Instead, they require coordinated action across education sectors, clearer national guidance, and sustained dialogue between universities, government, employers, and wider society. Policy must remain sensitive to disciplinary differences, equity and ethical considerations, and the need for experimentation within safe, regulated environments. Above all, AI should be approached not simply as a technical issue, but as a catalyst for rethinking the purposes and practices of education itself.

May 2026