Inclusive assessment design in the age of GenAI: Advancing pedagogy and practice
Stephanie McDonald1, Gary Saunders1, Ellie Kennedy1 and Alexander Tansell1
1 University of Nottingham, Nottingham, UK
In an educational landscape rapidly transformed by Generative AI (GenAI), inclusive assessment design is essential to sustaining equity and meaningful learning in higher education. This study took place at a Russell Group university in the United Kingdom. Focus groups were conducted with undergraduate and taught postgraduate learners studying science subjects, to investigate their experiences and perceptions of assessment practices in an increasingly AI-mediated learning environment. Findings show that learners value assessment approaches that centre human judgment, authenticity, meaningful choice, and critical engagement. Whilst recognising the potential for GenAI to support learning, they also expressed concerns about deskilling, inequitable access to GenAI tools, and assessment designs that do not facilitate meaningful learning. Rather than viewing GenAI solely as an academic integrity risk, learners emphasised the need for transparent guidance, stronger assessment literacy, and pedagogically informed decisions in assessment design. While GenAI is at the forefront of sector-wide discussions and planning in the context of curriculum development, our findings highlight the importance of grounding GenAI within broader principles of Universal Design for Learning and constructively-aligned assessment design – emphasising flexibility, transparency, and diversity – and the importance of actively involving learners as partners and co-creators in shaping inclusive assessment practices.
inclusive assessment, generative AI, human centred learning
Higher education in the UK is facing increasing political and public scrutiny, economic pressures, evolving learner expectations (Advance HE, 2025) and questions of legitimacy (Venning & Beech, 2026). Within this context, demonstrating educational excellence and value for money is essential for institutional sustainability (Advance HE, 2025). Inclusive assessment is central to this agenda. Grounded in Universal Design for Learning (UDL), it promotes “fair and effective assessment methods and practices that enable all learners to demonstrate to their full potential what they know, understand and can do” (Hockings, 2010, p. 34). Emphasising flexibility, choice, and multiple modes of representation and expression, inclusive assessment responds to the growing diversity of learners’ lived experiences, skills, backgrounds, and often competing demands on their time (Advance HE, 2025; Wright et al., 2025).
When proactively designed and constructively aligned, assessment can support deep learning rather than merely measure achievement (Boud & Falchikov, 2006; Boud & Molloy, 2013). Research highlights strategies such as diversifying assessment methods, authenticity, flexibility, transparency, and the development of assessment literacy and self-regulation as mechanisms for promoting equitable learning, reducing barriers, improving retention and progression, and addressing awarding gaps (Hanesworth, 2019). Such approaches not only reduce barriers to learning and limit the need for individual ad hoc accommodations for learners with diverse learning needs; they also function as good pedagogical practice for all learners (Bain, 2023; Hanesworth, 2019), thereby widening the opportunity for learners to demonstrate their learning and to support their professional development.
The rapid emergence of Generative AI (GenAI) has added further urgency to rethinking assessment design (Farrelly & Baker, 2023; Francis et al., 2025). Embedded within commonly used academic tools, such as pdf readers, word processing apps, spreadsheets, coding apps, and search engines, GenAI challenges assumptions about what assessments measure, how learning is demonstrated, and what constitutes learners’ work. This technology has prompted debates about the role and purpose of universities, what knowledge and competencies we expect graduates to develop, and what might be appropriately delegated to GenAI tools. Concerns include reduced learner engagement and critical thinking through overreliance on GenAI outputs (Bittle & El-Gayar, 2025), threats to the validity of traditional assessment methods evaluating the production of artefacts (e.g. essays, presentations, posters, blogs) rather than the learning process itself (Xia et al., 2024), and the limited reliability and potential bias of AI-detection tools (Farrelly & Baker, 2023). Equity issues are particularly significant. For example, detection tools may misclassify non-native English writing as AI generated, disproportionately affecting international and marginalised learners (Farrelly & Baker, 2023). Equity concerns extend beyond detection tools, with GenAI outputs being prone to algorithmic biases and the reproduction of systemic inequalities that risk going unnoticed and unchallenged (Hernández Nodarse et al., 2025). Further inequalities arise from differential digital competencies among learners (Francis et al., 2025).
At the same time, GenAI offers pedagogical opportunities aligned with inclusive assessment (Bittle & El-Gayar, 2025). It can function as an assistive tool, reducing cognitive and linguistic load for neurodivergent and disabled learners, and provide multilingualism support for learners for whom English is an additional language (Farrelly & Baker, 2023). GenAI can also provide immediate feedback on drafts of work and ideas (Arslan et al., 2024), support self-assessment and personalised learning (Xia, 2024) and adaptive assessments tailored to leaners’ needs and abilities (Arslan et al., 2024; Bittle & El-Gayar, 2025). Evidence also suggests that GenAI can positively impact learner motivation, confidence and engagement, particularly in relation to brainstorming ideas, clarifying understanding of challenging concepts, and helping to scaffold learning (Hmoud et al., 2024). These affordances position GenAI as a potential contributor to more inclusive and responsive design of assessment and learning.
Grounding the use of GenAI within UDL provides a pedagogically informed, inclusive, and learner-centred approach for harnessing these opportunities in the context of learning, teaching, and assessment. UDL advocates inclusive-by-design curriculum acknowledging learner diversity from the outset, rather than retrofitting courses after the design process (Meyer et al., 2013), offering flexibility across engagement, representation or how learning material is presented and accessed by learners, and action and expression or how learners demonstrate their learning (CAST, 2018; Meyer et al., 2013). However, realising these benefits requires attention to broader considerations including ethics, sustainability, and the potential risk of deskilling. Conceptualising GenAI as a learning partner, rather than a substitute for human judgment and knowledge creation, aligns with constructivist and social constructivist theories which emphasise active, socially mediated knowledge construction (Piaget & Inhelder, 1972; Vygotsky, 1978). This perspective foregrounds the importance of assessment designs that promote exploration, critical engagement, and reflection, supported by GenAI, but remaining human-led and human connected. While sector-wide policy statements and conceptual frameworks are developing, empirical research that centres learners’ perceptions and experiences of inclusive assessment within an evolving AI-mediated landscape remains limited (Bain, 2023). As institutions develop guidance on ethical AI use, embed AI literacy into curricula, and seek to maintain inclusive and robust assessment practices, understanding how learners perceive and navigate these changes is critical.
This study addresses this gap by utilising qualitative methodology to explore undergraduate and taught postgraduate learners’ perceptions and experiences of assessment in the context of emerging GenAI use, and to identify practices that are experienced as inclusive by learners. The research took place at a Russell Group university during the 2024/25 academic year. In line with other higher education institutions at the time, our university had not yet formulated a nuanced approach to GenAI use. The tendency was to prohibit the use of GenAI in formal assessments, unless otherwise specified, with potential cases of GenAI use considered within academic integrity processes. At the same time, GenAI was becoming increasingly embedded in tools used routinely by learners for reading, writing, coding and other academic activities. As its use by learners became more widespread, so did uncertainty about what constituted appropriate and inappropriate uses (Wright et al., 2025). The Russell Group had also recently issued principles on the use of generative AI in education (Russell Group, 2023), but these were not yet widely embedded across higher education. By situating GenAI within broader considerations of equity, flexibility, and support, the present investigation provides a holistic account of inclusive assessment at a time when institutions are rethinking their assessment strategies. In doing so, it offers timely evidence to inform practices that remain inclusive, equitable, and pedagogically sound in an AI-enabled higher education landscape.
A total of 19 students (17 undergraduate, 2 taught postgraduate) from across all departments in the Faculty of Science at the University of Nottingham participated in focus groups between March and April 2025, to explore their experiences and perceptions of assessment practices within their course. Subject disciplines in the faculty include Biosciences, Chemistry, Computer Science, Mathematical Sciences, Pharmacy, Physics and Astronomy, Psychology, and Natural Sciences. Three focus groups were conducted in total. Participants were recruited through student course representatives and student communications and were provided with an inconvenience allowance for participation in the study. Participants’ ages ranged from 18 – 24 years (M = 20.58, SD = 1.64). Participant demographic information can be found in Table 1.
Table 1. Participant demographic information
|
Gender |
Ethnicity |
Learning difference or disability |
Student status |
Discipline |
|
Female (47.4%) Male (52.6%) |
Asian (21%) Black (26%) Mixed (10.5%) Other (5.3%) White (36.8%) |
Presence of learning difference or disability (57.9%) None known (36.8%) Not reported (5.3%) |
UK (84.2%) International (15.8%) |
Biosciences (n=2) Chemistry (n=2) Computer Science (n=4) Mathematical Sciences (n=2) Pharmacy (n=3) Physics and Astronomy (n=3) Psychology (n=3) |
Ethical approval was granted by the School of Psychology Ethics Committee at the University of Nottingham (Reference: F1626). This study formed part of a larger project conducted in the Faculty of Science, in partnership with Advance HE, on developing inclusive learner experiences and building belonging (McDonald et al., 2025). Focus group questions were designed to capture learners’ experiences of assessment practices on their courses and perceptions on GenAI use in the context of assessment. Some of the questions addressed assessment more broadly, exploring perceptions and experiences around types of assessment, and diversity and choice in assessment methods. Others focused specifically on GenAI, exploring perceived benefits, concerns, and supporting mechanisms in utilising GenAI in the context of learning (see Appendix 1 for full set of questions). Rather than asking specific questions about links between inclusive approaches to assessment and GenAI, running the risk of leading students or closing off certain responses, we chose to leave the focus group questions more open, and instead looked for links between these two key topics at the analysis stage.
Focus group questions were co-created with two undergraduate students from the disciplines of mathematical sciences and natural sciences, contributing to the project as paid student consultants. Co-creation in research can promote learners’ feelings of belonging, confidence, and ownership over institutional decision-making and practice (Neary et al., 2014). The student consultants also facilitated the focus groups and transcribed focus group recordings. This practice, adopted in prior research (McDonald et al., 2025), helps reduce power dynamics and enhances authenticity, depth, and quality of the data by enabling participants to speak more openly with peers than with staff. Authors 1 and 4 were involved in the design of the study and the development of focus groups questions. Data analysis and interpretation of findings were conducted by Authors 1 and 2. Authors 1-3 contributed to the development of recommendations and the preparation of the manuscript. The research tool and its implementation were co-created with the student consultants as part of the larger project; however, this paper was not co-created and was written by the named authors.
Focus group discussions lasted between 60-90 minutes. Informed consent was obtained from participants prior to their participation in the focus groups. Discussions were audio recorded and transcribed verbatim for analysis.
Focus group transcripts were analysed by means of inductive thematic analysis, following the methodological procedure developed by Braun and Clarke (2006, 2012), to explore learners’ perceptions and experiences of assessment in the context of inclusive assessment design. The analysis was data-driven, with codes and themes developed through participants’ responses, rather than determined a priori. An essentialist/realist epistemological approach was adopted, with a focus on capturing participants’ experiences and the meanings of those experiences.
Reflexivity was a key aspect of the analytic process. As educators and researchers with experience and interests in inclusive teaching and learning practices and GenAI in the context of assessment, we recognise that our perspectives and prior experiences can shape aspects of data interpretation, including the ideas we attended to in the data, how themes were conceptualised and the inferences we made from them. We engaged in critical reflection and revisited the data throughout the analysis. This process helped in ensuring that the findings were derived directly from participants’ narratives, capturing key ideas in the dataset relevant to our research question.
Three themes were developed in the analysis, capturing learners’ experiences and perceptions of assessment practices: (1) Human-led learning in the age of AI: From threat to learning partner, (2) Equity challenges in assessment practices, and (3) Flexibility, diversity, and authenticity in inclusive assessment design. The themes are presented below, together with quotations from the focus group transcripts to evidence how the ideas encapsulated within each theme were manifested in the data. These themes reflect the central role that inclusive learning, teaching and assessment should play in considerations of GenAI use in higher education.
This theme captures learners’ perceptions of the role and utility of GenAI in learning in higher education. Focus group discussions revealed a nuanced view of GenAI, highlighting both its potential as a learning partner and concerns about its implications for learning.
Participants shared a range of experiences with GenAI in the context of learning – from no prior use to having used GenAI to support their learning. GenAI was viewed as a valuable learning tool that supports understanding of content, increases efficiency in completing learning activities (e.g., literature review), and supports assessment preparation. For example, one participant described its utility in streamlining learning activities: “I found it useful almost like a shortcut when doing research [...] when you use Google and you research stuff and it comes up with the AI overview [...] you can put [the references] into ChatGPT and ask them to [...] find the source and then you can look at the original paper [...] and then you can evaluate yourself, every one that you want to use as a source but it [...] makes it so much faster rather than having to scroll through everything [...]” (Participant 4, UG Year 3, Mathematical Sciences; FG3).
Whilst acknowledging its benefits for learners and the need for universities to stay current in an AI-infused age, participants also voiced concerns about overreliance on GenAI. They emphasised the importance of using GenAI as a tool that supports learning rather than enabling it to produce all the work for the learner. As one participant commented: “[...] I'm using it to make resources and then double checking that they're correct, and then you can use the resources. That's like a good way. But using it [...] [for example] I put the PowerPoint in. It's made notes for me now, I don't need to do anything. That's not the best way” (Participant 2, UG Year 2, Psychology; FG2). Participants further noted that reliance on GenAI may lead to passive learning habits. For example, one participant reflected on a shift in learning approaches as a result of GenAI: “solely relying on AI [...] would just make you lazy as a person [...] before this all started, we all used to do A Levels [...] we used to do work that actually properly learn and revise. But now with the introduction of AI [...] it’s our crutch [...] we just need it for everything [...] you need to [...] have a split as in use it for like revision, but don't solely rely on it because [...] [if used in this way] I'm not learning anything. And what's the point of my degree if I'm not learning anything?” (Participant 6, UG Year 1, Physics and Astronomy; FG2). This distinction was seen as critical in maintaining a human-centred approach to learning.
While participants acknowledged that integration of GenAI in teaching and learning more broadly can support development of AI literacy, they remained cautious about its role in the context of assessment. Participants were particularly concerned about the diminishing role of human agency in learning, especially when AI is either formally integrated into assessment design or used by learners to support them in completing assessment tasks. Participants highlighted the potential of a negative impact on learning: e.g., “[...] what you're doing is just copy and pasting or the robot gives you you're not really learning anything” (Participant 6, UG Year 1, Physics and Astronomy; FG2) and the risk of deskilling, with the potential of undermining the development of essential discipline-specific skills. As one participant commented: “before ChatGPT, I would code and understand what I'm doing, but sometimes it's just easy, now that ChatGPT is about to just put the question into ChatGPT” (Participant 3, UG Year 4, Computer Science; FG1). Participant discussions further revealed the value of restricting the use of GenAI for tasks which are associated with discipline-specific competencies required in professional contexts. For example, one participant noted: “if you are going to get a coding job [...] they’re going to go back to asking you to do live coding in front of them [...]. So, I think having assessment for certain things where you can use AI [...]” (Participant 3, UG Year 2, Physics and Astronomy; FG2). Participants commented that assessment methods, such as those that are coursework based, are perceived to be more vulnerable to GenAI use, in comparison to assessments conducted in controlled settings, such as in-person closed-book examinations: e.g., "having closed book exams for [certain types of tasks] is probably the only way to make sure people don’t actually use AI” (Participant 3, UG Year 2, Physics and Astronomy; FG2).
Academic integrity was identified as a key concern in the age of GenAI. Participants emphasised the need to distinguish between using GenAI to support learning and using it to generate assessed work. The latter raised ethical questions and concerns about who is truly being assessed, the learner or the AI tool. As one participant mentioned: “[...] you can use it to prompt yourself [...] to help come up with ideas and it's fine. But if it's like [...] you can use it to help answer questions [...] am I really answering the questions anymore? Are they now assessing me or the AI?” (Participant 4, UG Year 1, Biosciences; FG2).
Others equated the use of GenAI in assessment with cheating, noting a shift in institutional guidance on its use for assessments but maintaining a critical stance on its role in the learning process, e.g., “it's not exactly using your brain, it's just relying on something else to come up with it [...] throughout like the whole of university, we've always been told you can't use this [GenAI] [...] and then obviously because technology is evolving [...] we're now allowed to [...] use it [...] it's always been like an atmosphere of you can't use it [...] if it's doing it for you, then that's not your work anymore [...] it's theirs” (Participant 1, UG Year 4, Biosciences; FG1).
Participants expressed concerns about the reliability of GenAI outputs, due to the potential of inaccurate information, which was seen as particularly prominent in some subject areas (e.g., mathematics, physics, calculations), placing the learner at risk should outputs not be evaluated and cross-checked. For example, one participant commented: “I feel like it can get things a lot more wrong than you'd have thought it could have done. So don't fully trust it [...] [in relation to] maths, physics type questions” (Participant 5, UG Year 1, Mathematical Sciences; FG1). Participants expressed more trust in AI when GenAI tools were tailored to the discipline, e.g., “if the AI [is] purpose built for the subjects that we are doing, then I’d be happier to use it and trust it more” (Participant 5, UG Year 1, Chemistry; FG3) and when learners were guided to use AI in subject- and task-specific ways that were designed to support, rather than replace, learning. For example, one participant commented: “if there’s an AI software that’s going to be used for an assessment, it definitely needs to have its limits and the type of information it can give you [...] for my lab project this year, it was actually using AI notebook [which was programmed by researchers within the department] [...] it gave you the numbers, but it was up to you to [...] then figure out how to make sense of the numbers. It didn’t then generate a whole report for you [...] if I was to put the numbers I got in ChatGPT, it could generate and write my entire report but with the AI notebook we had, it just gave you the numbers and it was in your hands to then actually make up your report and write and explain what these numbers mean. I feel like the limitations and the information that the software can give you [...] support the learning” (Participant 1, UG Year 3, Chemistry; FG3), highlighting how programming GenAI tools for the context of a particular subject or an assessment task can facilitate critical thinking and human-led learning.
Focus group discussions highlighted that to support learners in making informed choices and effective, critical uses of GenAI in their learning, GenAI needs to be purposefully integrated in teaching and learning, including assessment design. This includes the need to design supporting mechanisms to enable learners to develop their AI literacy skills and to support an institutional transition from forbidden use to acceptable, or even encouraged, use of GenAI in higher education. One participant noted: “I think embracing it [AI] is the only way forward because like even if the lecturer doesn't say everyone use ChatGPT in this assessment, they're going to do it anyway. So, I feel like encouraging it and enforcing shift of guidelines around it will probably be a better approach” (Participant 1, FG2), whilst another commented that “[...] guidelines don't necessarily need to just be about restricting behaviour, it can also be about best use and best practice” (Participant 1, UG Year 3, Computer Science; FG2). Embedding critical AI literacy into teaching and learning in ways that both align with particular assessments and support the development of learner-specific awareness of learning processes can help to address a potential lack of understanding in AI capabilities, e.g., “[...] if you give it a question, it's going to give everyone the same answer or relatively similar answers. I feel like you'll kind of limit the range of expression [...] (Participant 1, FG3) and further support learners’ assessment literacy (e.g., “it's just really helpful when [...] you get the chance to actually learn about what it's like the data is giving you rather than just [...] copying and pasting it [...] the workshops [...] we had [supporting this understanding] are helpful” (Participant 1, UG Year 3, Chemistry; FG3).
This theme captures participants’ perspectives and experiences around equity in assessment design, in light of evolving technologies and diverse learner needs.
Participants expressed concerns that embedding tools such as GenAI into assessment tasks could exacerbate existing inequalities among learners, especially where variability in AI literacy and access to technology exist. For example, one participant highlighted perceived unfairness of requiring AI use in assessments, noting: “I feel like having it as part of an assessment would be a bit unfair because not everyone has been taught how to use it properly. And some people don't have the [...] technology skills to work with it properly [...] that could be an issue” (Participant 6, UG Year 2, Physics & Astronomy; FG3). Participants also expressed feeling the pressure of using GenAI, as there is an expectation that their peers would be using it, putting them at a disadvantage if they choose not to (e.g., “I wouldn't want it to be like you have to use it in an assessment because [...] I don't really trust that it gets it right [...] but then I wouldn't want to not use it knowing that everyone else is using it and potentially benefiting from using it” (Participant 4, UG Year 3; Mathematical Sciences; FG3). The implications of unequal usage of GenAI were particularly prominent in the context of group work, where this could lead to an imbalance in workload among learners. For example, one participant expressed: “[it] becomes even more problematic when it's a group project and you have somebody so clearly using AI and the rest of them just aren't, especially in coding [...] and [others] were just doing the majority of the work [...] (Participant 3, UG Year 2, Physics & Astronomy; FG2).
Concerns also extended to the availability of different versions of AI tools, with participants pointing out that access to paid versions by some learners could introduce bias. As one participant commented: “[...] in this context where you're required to use AI […] there are paid versions of things like ChatGPT. So, someone might be using the free version, someone might be using the paid version [...] and that’s just a bit biased [...] unfair” (Participant 3, UG Year 4, Computer Science; FG1). These disparities were seen as undermining the principle of equal opportunity in assessment. Beyond emerging technologies, discussions also focused on more traditional assessment formats, such as closed book time limited examinations, which were viewed as largely inequitable. These were described as focusing solely on memory, rather than providing opportunities for learners to demonstrate their learning and capabilities. One participant reflected: “[...] in the actual field, do I have to remember everything? No, but in a close book exam, I need to remember everything and that's just unfair [...] that's not their ability, that's just whether their brain can deal with it [...] there's got to be some fairness there (Participant 1, UG Year 4, Biosciences; FG1).
There was strong support for assessment formats that allow learners to demonstrate their strengths, such as coursework and remote, open-book assessments, where learners are able to control their environment, as well as varied assessment types, supporting diverse learner needs. For example, one participant commented: “[...] different people have different ways of being able to express their knowledge [...] having a variety of different things allows people to maybe do better in some areas than others, which I think is makes it more fair [...]” (Participant 2, UG Year 2, Psychology; FG2).
Participants also emphasised the importance of transparent and detailed marking criteria to support assessment literacy. While marking rubrics were generally seen as helpful, learners noted that their effectiveness depended on clarity of expectations and the ability to link feedback to the criteria. For example, one participant mentioned: “[...] having a very detailed rubric and to see how it's marked [...] does really help because I find that when I'm doing like essays, I write [...] alongside of criteria and use [this as a] sort of check box" (Participant 1, UG Year 3, Chemistry; FG3). Formative assessments and opportunities to engage with marking criteria were viewed as key learning activities in building confidence and assessment literacy, e.g., “for one of my coursework essays, I could submit a shortened version [...] that lecturers would give you feedback on which I found quite helpful because you can see where you’ve got to improve” (Participant 4, UG Year 1, Biosciences; FG2). Participant responses also allude to the potential for GenAI to support assessment for learning, particularly in the form of formative tasks. For example, one participant commented: “in Pharmacy [...] [an AI tool is used in teaching and learning to support] practising consultations and preparing for OSCE exams [...] it was really good” (Participant 2, UG Year 1, Pharmacy; FG3).
This theme encapsulates perspectives on diversity of assessment, flexibility in assessment, and authenticity as identified by learners.
Focus group discussions highlighted the value of diverse assessment methods in fostering engagement, supporting deeper learning, and the development of a broad range of competencies. For example, one participant commented: “There is positive doing multiple different types of assessments in the fact that it will teach you like a range of different work like skills, which will be probably helpful for future life situations [...]” (Participant 4, UG Year 1, Biosciences; FG2). On the other hand, a lack of diversity in assessment was described as limiting learners’ breadth of learning. For example, repeated use of similar assessment formats (e.g., examinations) led learners to focus narrowly on predictable content, potentially creating gaps in knowledge and understanding of the broader curriculum, e.g., “[...] the past papers are basically exact same year after year [...] that really does help me narrow down one of them to study [...] because we kind of have a lack of diversity, I guess, it kind of limits my overall knowledge [...]” (Participant 6, UG Year 2, Physics and Astronomy; FG3).
However, adapting to different types of assessment can be challenging. Participants spoke about the challenges in preparing for multiple assessment formats simultaneously or throughout the course of their degree. For example, one participant expressed: “It's challenging to adapt [...] change for some of us is very difficult. So having to change to different things [...] you've got to try and think about how you're going to do both things at the same time” (Participant 1, UG Year 4; Biosciences; FG1).
Participants valued having a choice in assessment formats, as this was viewed as empowering and supportive of autonomy in learning. The ability to select assessment tasks that aligned with personal strengths and professional practice was seen as supporting motivation and quality in the work that was produced. For example, one participant expressed: “[I was] allowed to choose what actually will help me communicate better and actually take some transferable skills into the workplace [...]” (Participant 1, UG Year 4; Biosciences; FG1). Choice was also seen as beneficial for learners with learning differences, enabling a more equitable learning experience, e.g., “[...] when you have the choice [...] there's like the benefits of not feeling shortcomed by your assessment [...] but still being able to get the same experience that everyone else is having” (Participant 1, UG Year 3, Chemistry; FG3). However, participants also recognised that choice must be embedded meaningfully within curriculum design to enable the development of a broad range of competencies. As one participant noted: “I like choice too, but I think you should have to do one of every exam, maybe cause I don't think you should be able to get by your entire degree without writing [...] without doing coursework or without doing an exam [...] both give you a different set of skills” (Participant 3, UG Year 4, Computer Science; FG1), highlighting that flexibility in assessment can be beneficial to enable a diverse range of learners to succeed, but that flexibility at individual module assessments should be carefully planned with an eye to programme-wide assessment diet and programme learning outcomes.
Whilst acknowledging the benefits, participants also expressed concerns about the perceived equivalence of different assessment options. There was a sense that certain formats would be marked more favourably or allow for greater creativity, leading to strategic decision-making and uncertainty. For example, one participant commented: “We had one choice where you could record a video or do an essay, which I personally liked, but a lot of my peers were afraid that they would get marked worse if they did the essay” (Participant 3, UG Year 4, Computer Science; FG1). Others noted that choice could introduce ambiguity around marking criteria, e.g., “[...] sometimes choice comes with a lack of clarity. And especially for example, if it's an assessment where you have to choose your own topic or create something from your own ideas or knowledge. In terms of how that would be marked or assessed can sometimes be a bit unclear because you're the one kind of creating rules. And so that might be the only challenge” (Participant 1, UG Year 3, Computer Science; FG2). These findings suggest that, where flexibility of assessment is offered, learners will need to be supported in undertaking their chosen method of assessment.
Authentic assessment was viewed as particularly valuable, aligning with professional practice. Participants spoke favourably of assessment types that reflected real-world tasks, e.g., “[...] OSCE exam and [...] group presentation [...] real life practise as a pharmacist [...] it's important that you know how to consult a patient [...] and [...] working in groups [...] with other healthcare professionals [...]” (Participant 2, UG Year 1, Pharmacy; FG3). Other formats were not viewed as directly relating to course content or professional practice, e.g., “I just don't see how portfolios and presentations have been useful to me at all [...]” (Participant 5, UG Year 1, Mathematical Sciences; FG1), reflecting the need for greater transparency in the rationale behind assessment design, and in particular its relevance to learning outcomes and employability.
This study has explored learners’ experiences and perspectives about inclusive assessment at a time when GenAI is becoming increasingly used by learners and embedded within tools and software used for learning (Freeman, 2025). The findings highlight the importance of inclusive assessment design as a foundational principle in a rapidly shifting technological landscape and the increasing capabilities of GenAI. Within this context and drawing on the findings and our expertise as educators, consideration is given to the potential of embedding inclusive design principles into a human-centred approach to assessment that utilises GenAI as a learning partner but with human critical oversight over if, when and how GenAI is used in the learning process.
Focus group discussions revealed a range of barriers which might inhibit learner success in assessment; these align with barriers reported elsewhere in the literature, such as experiencing cognitive overload arising from the volume and complexity of assessed tasks (Sweller, 1988). Other barriers commonly experienced by learners in contemporary higher education include compressed submission deadlines, and competing academic, personal, and financial pressures; while these were not specifically mentioned by the participants, they are considered key features of the current context (Wright et al., 2025). Learners commented how barriers were exacerbated by unclear and inconsistently applied marking criteria (O’Donovan et al., 2004), the use of assessment formats without support for assessment literacy, and differential use and understanding of assessment formats (Biggs, 1999). These insights highlight the foundational importance of embedding inclusive design principles consistently at course level, module design, and learning and teaching strategies (McDonald et al., 2025) and that there is still much work to do to ensure inclusivity in course design and delivery.
The growing prevalence of GenAI has added further complexity to existing issues around inclusivity. Much of the literature prior to 2024 tends to frame GenAI either as a threat to academic integrity or a neutral technological tool to be integrated to enhance learning and teaching (Yusuf et al., 2024). Participants in this study, however, reflected a more nuanced and, in some respects, more troubling set of concerns, consistent with recent sector-wide research (Attewell, 2025; Digital Education Council, 2026; Freeman 2025). The participants’ primary concern was not solely about the misuse of GenAI but something more foundational – the capacity of GenAI to disrupt the meaning of learning and assessment in higher education (Freeman, 2026). They were particularly concerned with the potential of GenAI to diminish the role of human agency in learning, especially where it was possible to outsource learning to the technology. These concerns are echoed in sector-wide studies on learners’ attitudes to AI in higher education (Dickinson & Marshall, 2026; Stephenson & Armstrong, 2026).
In response to this uncertainty, learners consistently foregrounded the importance of critical human agency. Where learners acknowledged that GenAI could usefully support research, aid comprehension, or assist with revision, they made a clear conceptual distinction between using GenAI to scaffold learning and using it as a substitute for learning (Chan & Hu, 2023). This distinction closely aligns with constructivist perspectives, which emphasise learning through active experience and reflection rather than in the passive reception of information (Piaget & Inhelder, 1972). From this perspective, meaningful engagement with GenAI depends on its integration with authentic learning and assessment activities that necessitate by design reflection, critical evaluation, and personal sense-making rather than passively producing information and uncritically accepting and presenting that information for assessments (Su et al., 2026). Dickinson and Marshall (2026) foreground a link between human agency and assessment design, where their findings indicated that students use AI very differently when they know that an “accountability moment” is part of a forthcoming assessment (p. 19). Another factor making human-led critical approaches necessary is AI sycophancy; GenAI has a designed-in tendency to tell users what they want to hear in a coherent way that may be agreeable but not necessarily factual (Turner & Eisikovitis, 2026).
The application of a constructivist perspective can be further extended through Vygotsky’s (1978) conceptualisation of learning as an inherently social process that requires interaction with more knowledgeable others. Understanding learning in these terms, GenAI could be used as a learning partner that enriches rather than replaces the dialogical and social aspects of learning. This addresses some of the learners’ concerns and creates a meaningful place in the learning process for both learners and educators as well as promoting a collective critical human element. Without such grounding, the risk is that learners become isolated passive consumers of GenAI content rather than active and collaborative producers of knowledge. This concern underlines the importance of embedding appropriate pedagogical approaches when using GenAI for learning to preserve the human-in-the-loop (Askari, 2025; Guo et al., 2024).
Participants also discussed issues around academic integrity (Yusuf et al., 2024). They demonstrated an awareness of the distinction between what they considered to be legitimate and illegitimate uses of GenAI, but argued for more consistent and explicit institutional guidance, clearer boundaries, and greater transparency about what is and is not acceptable use (Freeman, 2025). The absence of such guidance does not prevent or deter learners from using GenAI but rather leaves them to navigate its use within an uncertain landscape of informal use and inconsistent practice. This can create conditions that risk normalising uncritical and unreflective dependency on GenAI (Johnston et al., 2024), rather than a critical, informed, and human-led approach, as advocated by the learners in the focus groups. To ensure this, it will be important to support learners to be transparent about how and when they have used GenAI both to ensure critical engagement with the technology but also to ensure that they have followed the guidance on GenAI use (de Almeida McLoughlin et al., 2025). As institutions move away from blanket prohibitions towards providing detailed guidance on AI use (Dickinson & Marshall, 2026; Stephenson & Armstrong, 2026) it will be important to bear these considerations in mind.
Some participants found that coursework was more susceptible to GenAI use than other forms of assessment, in that it was easier to put a coursework assessment task into GenAI. Participants did consider exams as a secure assessment format but considered that this approach to assessment was unfair and did not allow them to fully demonstrate their learning and their capabilities. While participants were unclear about what forms of assessment would be most appropriate, it is possible to draw from their discussion that where assessment is intended to evaluate learning, its design must engage with the learning process itself and focus on contextual application, reflection, and the exercise of critical human judgment even in contexts where GenAI use is permitted (Su et al., 2026). This approach entails building assessment literacy into learning and teaching wherein assessments are scaffolded and formative feedback is provided at different stages. This allows for greater engagement with the learning process and how it is progressing rather than focusing on the final output. Also, more dialogical approaches to assessment that allow learners to explain and demonstrate their learning are useful ways to approach assessment in the age of GenAI (Maregere, 2025). Furthermore, participants expressed an interest in authentic assessment that allowed them to apply their learning to real or simulated situations and thereby develop and demonstrate key competencies. These approaches can support a greater level of reflection and metacognition about the learning process that not only supports an effective learning environment (Phenwan, 2016) but also reduces the likelihood that learners will outsource their learning to GenAI (Combrinck & Loubser, 2025).
Concerns about equitable use of GenAI in assessment were also evident in focus group discussions. Rather than levelling academic playing fields, learners discussed how GenAI risks amplifying existing inequalities by privileging those with greater digital capital, including familiarity with AI tools, access to reliable technology, and the means to subscribe to premium platforms that outperform their free equivalents (Freeman, 2025; Rahman & Saunders, 2024). This introduces a resource-based advantage that sits uncomfortably alongside principles of fair and inclusive assessment and highlights the importance of building digital literacy across all learner groups (Advance HE, 2025) and the institutional responsibility to ensure equitable access to technology (Freeman, 2025). This response will also need to take account of emerging gender and subject differences in AI use, where studies have shown that male learners and learners studying STEM subjects more likely to use GenAI (Dickinson & Marshall, 2026). Without these interventions, GenAI has the potential to become another mechanism through which existing structural inequalities are reproduced or exacerbated (Rahman & Saunders, 2024).
One way of responding to these tensions is offered by Universal Design for Learning (UDL). Focus group participants spoke positively about having meaningful choice in assessment, including flexibility in format and topic and how this can be empowering and motivating. This is because it would allow them to draw on their strengths, manage anxiety, and demonstrate learning in ways that align with their lived experiences, identities, and aspirations. These findings strongly support the principles for UDL which advocates for the provision of multiple pathways to demonstrate learning that enhances inclusivity and learner agency without diminishing academic standards (Meyer et al., 2013). Nevertheless, flexibility should not be seen as the only or the primary approach to inclusivity, including where GenAI is concerned. To ensure that courses are constructively aligned to learning outcomes, it is important that opportunities to develop skills and knowledge are located within different modules and assessments (Biggs, 1999). This means that some modules will develop AI literacy that is then assessed and other modules will focus on skills and knowledge that do not use GenAI. This entails strategically redesigning aspects of learning and assessment to support learners to learn and demonstrate learning in the age of AI, while keeping humans at the centre.
Emerging concepts notwithstanding, educators can take comfort from the fact that many elements already known to support inclusive learning and assessment are likely to support learners in the age of AI. For example, the following strategies are becoming more important than ever: a) ensuring that all course learning outcomes are assessed; b) giving learners with different strengths the opportunities to use them; c) planning skills development (e.g. written and verbal communication, group working, professional skills) across a course’s years of study. Well-established concepts such as constructive alignment (Biggs, 1999) can be deployed when offering multiple assessment options so that there is equivalence in rigour, shared evaluative criteria, and transparent communication so that choice promotes inclusion rather than creating new anxieties. These same concepts can be applied when planning learning and assessment to take account of GenAI.
Overall, our findings suggest that learners value assessments that are authentic, diverse, and prepare them for future employment. Based on the findings, it is possible to conclude that assessments requiring contextual understanding, personal reflection, and evaluative judgment have the potential to be more meaningful, motivating, and less susceptible to misuse and replacement of human knowledge production by GenAI outputs. This is because those types of assessment demand critical human-led engagement and represent a pedagogically robust response to the challenges identified in this research. We can ground GenAI use within established principles of inclusivity and authenticity. These principles and approaches can offer a constructive way forward that positions GenAI not as a threat to academic integrity that needs to be policed or uncritically adopted, but as a learning partner whose value depends entirely on the quality of the critical human engagement by learners and educators in the learning and assessment processes. Given the concerns and insights provided by participants, it is important that learners are involved in discussions about if, when and how GenAI is used in learning and teaching, as partners and co-creators of learning design.
Drawing on the research findings and the discussion of those findings, we offer a series of recommendations that can be implemented at three different levels.
· Institutions should develop a clear, coherent stance on GenAI which includes the role of AI in learning and the centrality of critical human engagement in the process. Such a stance should be aligned to teaching and learning strategies, to support institutions in moving forward with GenAI, from blanket prohibition towards transparent, constructive integration in learning. It is also important that strategies provide autonomy for courses to develop discipline appropriate responses.
· Institutional policies should set expectations which normalise provision of and engagement with criteria or marking rubrics, and formative assessments.
· Institutions should update their digital literacy frameworks to include AI-specific competencies and critical human engagement.
· To support learner digital literacy, educator development is essential for building capability and pedagogical understanding of effective and appropriate use (and non-use) of GenAI in teaching and learning.
· To support equity of access, institutional policies should incorporate clear guidance which ensures GenAI use in assessments does not require resources which may not be available to all learners. This includes providing institutionally licensed AI platforms to counter inequities between free and paid versions. Institutions can further proactively support the development of localised AI tools, closed systems, and discipline specific interfaces.
· Given the richness of the discussions about their experiences and perspectives of assessment and GenAI, it is important that learners are included in the design of learning, teaching, and assessment. While not stated by participants, the authors have argued previously that this can be done most meaningfully through co-creation, which can support critical engagement, inclusivity, and sense of belonging (McDonald et al., 2025).
· Courses should consider providing a strategically planned, diverse range of assessment types that are constructively aligned and offer an appropriate balance of GenAI use and non-AI use across the course, to support an appropriate range of discipline-specific professional competencies (QAA, 2023; TESQA, 2023).
· Well-designed, constructively aligned assessment choice can enhance inclusivity, but options must remain equivalent and clearly communicated and explained to learners.
· Building assessment literacy at course level, including in relation to the use of GenAI, through curriculum activities that complement institutional support, can support human-led learning and skills development.
· Providing clear expectations regarding permitted tools, ensuring all learners have access to these, is essential.
· This can further be supported as part of the assessment design by asking learners to evidence the use of tools and critically discussing any role played by GenAI in the development of the piece of work.
· Supporting learners to develop critical, context-specific use of GenAI alongside AI-free methods where these are pedagogically important.
· Clarity of expectations can be facilitated through engagement with assessment criteria and how these link to learning outcomes and feedback .
· Shared understanding can also be promoted by providing explicit GenAI guidance for assessment activities, including acceptable and/or non-permitted uses of GenAI, recommended tools or approaches, and clear expectations for transparency in how GenAI has (or has not) contributed to learner work, aligned with learning outcomes. Strategies to address transparency can vary and may include providing prompts used; integrating and signposting AI-generated elements as part of the assessment task; or submitting a critical commentary on how AI use contributed to the work.
· Learning activities should support learners to view and use GenAI as a learning partner rather than outsourcing their learning to it.
The findings are based on focus groups at a single institution within one faculty. However, participants were learners across all schools within the Faculty of Science, thus, ensuring representation across a range of different disciplines. The learners’ accounts of GenAI use are self-reported and may reflect the experiences of a small group of self-selecting individuals. The findings of our research are not dissimilar to recent research conducted on the use of GenAI, although the prevalence of its use seems to be much higher now with 94% of students stating that they have used GenAI to help with assessed work (Stephenson & Armstrong, 2026). This has substantially changed assessment in higher education with institutions trying to adapt to the new technology (Stephenson & Armstrong, 2026), which was at an early stage of development in our research. Interestingly, almost half stated that GenAI has improved the student experience (Stephenson & Armstrong, 2026). Sixty-eight per cent believe that AI skills are essential, but only 48% feel that staff are helping them to develop these skills (Stephenson & Armstrong, 2026). These last three points are similar to the findings of our research and show that our findings are timely and echo those in research conducted on larger scale. It should also be noted that participants in our study experienced a large-scale prohibition of GenAI use in their assessments at a time when GenAI tools were becoming widely available but teaching of critical AI literacy was not yet widespread. The findings in this study therefore provide a useful baseline against which to read emerging research in the sector. Future studies would benefit from exploring perspectives of learners who are experiencing a more nuanced approach to AI guidance and with a wide range of experiences in the use of GenAI.
The research contributes to the literature by offering a learner-centred conceptualisation of inclusive assessment in the age of GenAI. It highlights that learners are not simply passive recipients of technological change but provide critical insight on their own learning experiences. The learners’ discussions highlight the need for inclusive assessment in the age of GenAI, which is transparent, authentic, pedagogically informed, and constructive aligned. Given the insights provided by learners, it will likely be beneficial to involve them meaningfully as co-creators in assessment design as GenAI is further integrated into inclusive assessment practices.
We would like to thank Bethan Williams and Nicola Howie for their contribution to this work, including input into the development of focus group questions and facilitating the focus groups.
The authors have no relevant financial or non-financial competing interests to report.
Advance HE. (2025). Securing educational excellence in higher education at a time of change. Advance HE. https://www.advance-he.ac.uk/knowledge-hub/securing-educational-excellence-higher-education-time-change
Arslan, B., Lehman, B., Tenison, C., Sparks, J.R., López, A.A., Gu, L., & Zapata-Rivera, D. (2024). Opportunities and challenges of using generative AI to personalize educational assessment Frontiers in Artificial Intelligence, 7:1460651. https://doi.org/10.3389/frai.2024.1460651
Askari, M. (2025). Human-in-the-loop: A faculty-led model for integrating GenAI in executive economics instruction. The International Journal of Management Education, 23(3). https://doi.org/10.1016/j.ijme.2025.101247
Attewell, S. (2025, May 22). Student perceptions of AI 2025. Jisc. https://www.jisc.ac.uk/reports/student-perceptions-of-ai-2025
Bain, K. (2023). Inclusive assessment in higher education: what does the literature tells us on how to define and design inclusive assessments? Journal of Learning Development in Higher Education, (27). https://doi.org/10.47408/jldhe.vi27.1014
Biggs, J. (1999). Teaching for quality learning at university. SRHE and Open University Press.
Bittle, K., & El-Gayar, O. (2025). Generative AI and academic integrity in higher education: A systematic review and research agenda. Information, 16(4), 296. https://doi.org/10.3390/info16040296
Boud, D., & Falchikov, N. (2006). Aligning assessment with long‐term learning. Assessment & Evaluation in Higher Education, 31(4), 399–413. https://doi.org/10.1080/02602930600679050
Boud, D., & Molloy, E. (2013). Rethinking models of feedback for learning: The challenge of design. Assessment & Evaluation in Higher Education, 38(6), 698–712. https://doi.org/10.1080/02602938.2012.691462
Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101. https://doi.org/10.1191/1478088706qp063oa
Braun, V., & Clarke, V. (2012) Thematic analysis. In H. Cooper, P. M. Camic, D. L. Long, A. T. Panter, D. Rindskopf, & K. J. Sher (Eds.), APA handbook of research methods in psychology, Vol. 2: Research designs: Quantitative, qualitative, neuropsychological, and biological (pp. 57-71). American Psychological Association.
CAST. (2018). UDL Guidelines Version 2.2. https://udlguidelines.cast.org/
Chan, C.K.Y., & Hu, W. (2023). Students’ voices on generative AI: perceptions, benefits, and challenges in higher education. International Journal of Educational Technology in Higher Education, 20(43). https://doi.org/10.1186/s41239-023-00411-8
Combrinck, C., & Loubser, N. (2025). Student self-reflection as a tool for managing GenAI use in large class assessment. Discover Education, 4(72). https://doi.org/10.1007/s44217-025-00461-2
de Almeida McLoughlin, M., Lateef, B., Peirce, E., Collins, L., Parkinson, S., Taggart, J., Tansey, E. and Keane, A. (2025). Ethical AI Toolkit: A student resource to generative AI. Advance HE. https://www.advance-he.ac.uk/news-and-views/ethical-ai-toolkit-student-resource-generative-ai
Dickinson, J., & Marshall, M. (2026). Trained to stop learning? How students are experiencing assessment and learning in the age of AI. Wonkhe. https://wonkhe.com/wp-content/wonkhe-uploads/2026/03/Trained-to-stop-learning-F.pdf
Digital Education Council. (2026). Student voices on AI: An actionable guide for institutions and faculty. https://www.digitaleducationcouncil.com/post/student-voices-on-ai-an-actionable-guide-for-institutions-and-faculty
Farrelly, T., & Baker, N. (2023). Generative artificial intelligence: Implications and considerations for higher education practice. Education sciences, 13(11), 1109. https://doi.org/10.3390/educsci13111109
Francis, N.J., Jones, S., & Smith, D.P. (2025). Generative AI in Higher Education: Balancing Innovation and integrity. British Journal of Biomedical Science, 81, 14048. https://doi.org/10.3389/bjbs.2024.14048
Freeman, J. (2026). Learner generative artificial intelligence survey 2026. Higher Education Policy Institute. https://www.hepi.ac.uk/wp-content/uploads/2026/03/HEPI-Report-199-Gen-AI-Survey-2026.pdf
Freeman, J. (2025). Student generative artificial intelligence survey 2025. Higher Education Policy Institute. https://www.hepi.ac.uk/reports/student-generative-ai-survey-2025/
Guo, H., Yi, W., & Liu, K. (2024). Enhancing constructivist learning: The role of generative AI in personalised learning experiences. In Proceedings of the 26th International Conference on Enterprise Information Systems – Volume 1 (p. 767–770). https://doi.org/10.5220/0012688700003690
Hanesworth, P. (2019). Inclusive assessment: Where next? Advance HE.
Hernández Nodarse, M., Fonseca Torres, W., Ponce de León, D., Villarroel Henríquez, V., & López, A.R. (2025). Ethics, generative artificial intelligence, and educational assessment: An analysis of university students’ perceptions. Salud, Ciencia y Tecnología, 5, 2359. https://doi.org/10.56294/saludcyt20252359
Hmoud, M., Swaity, H., Hamad, N., Karram, O., & Daher, W. (2024). Higher education students’ task motivation in the generative artificial intelligence context: The case of ChatGPT. Information, 15(1), 33. https://doi.org/10.3390/info15010033
Hockings, C. (2010). Inclusive learning and teaching in higher education: A synthesis of research. Higher Education Academy.
Johnston, H., Wells, R. F., Shanks, E. M., Boey, T., & Parsons, B. N. (2024). Student perspectives on the use of generative artificial intelligence technologies in higher education. International Journal for Educational Integrity, 20(2). https://doi.org/10.1007/s40979-024-00149-4
Maregere, B. (2025, October 30). Dialogic assessments are the missing piece in contemporary assessment debates. Wonkhe. https://wonkhe.com/blogs/dialogic-assessments-are-the-missing-piece-in-contemporary-assessment-debates/
Meyer, A., Rose, D.H., & Gordon, D. (2013). Universal design for learning: Theory and practice. CAST Professional Publishing.
McDonald, S., Saunders, G., Eley, A., Parish, G., Lister, K., Tansell, A., Reid, K.L., Morgan, J., & Carver, L. (2025). Student perspectives on building belonging through inclusive teaching and learning practices and learning opportunities. Open Scholarship of Teaching and Learning, 5(1). https://doi.org/10.56230/osotl.137
Neary, M., Saunders, G., Hagyard, A., & Derricott, D (2014). Student as producer: Research-engaged teaching, an institutional strategy (Project Report). Advance HE. https://repository.lincoln.ac.uk/articles/report/Student_as_producer_research-engaged_teaching_an_institutional_strategy/24337309
O’Donovan, B., Price, M., & Rust, C. (2004). Know what I mean? Enhancing student understanding of assessment standards and criteria. Teaching in Higher Education, 9(3), 325–335. https://doi.org/10.1080/1356251042000216642
Phenwan, T. (2016). Teaching of reflection in higher education: A narrative review. MedEdPublish. 4(53). https://doi.org/10.12688/mep.20389.1
Piaget, J., & Inhelder, B. (1972). The psychology of the child. Basic Books
Quality Assurance Agency (2023). Reconsidering assessment for the ChatGPT era: QAA advice on developing sustainable assessment strategies. https://www.qaa.ac.uk/docs/qaa/members/reconsidering-assessment-for-the-chat-gpt-era.pdf
Rahman, N., & Saunders, G. (2024, September 11). Access to premium AI services is a significant concern for learners. Wonkhe. https://wonkhe.com/blogs/access-to-premium-ai-services-is-a-significant-concern-for-students/
Russell Group. (2023). Principles on the use of generative AI in education. https://www.russellgroup.ac.uk/policy/policy-briefings/principles-use-generative-ai-tools-education
Stephenson, R., & Armstrong, C. (2026). Student generative AI survey 2026. HEPI/Kortext. https://www.hepi.ac.uk/reports/student-generative-ai-survey-2026/
Su, F., O’Dea, M., Wood, M., & Seidl, E. (2026). Embracing a new world: Authentic assessment designs in an age of generative artificial intelligence (GenAI). Perspectives: Policy and Practice in Higher Education, 1–12. https://doi.org/10.1080/13603108.2025.2601741
Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257-285. https://doi.org/10.1016/0364-0213(88)90023-7
Tertiary Education Quality and Standards Agency (TESQA). (2023). Assessment reform for the age of artificial intelligence. https://www.teqsa.gov.au/guides-resources/resources/corporate-publications/assessment-reform-age-artificial-intelligence
Turner, C., & Eisikovits, N. (2026). Programmed to please: the moral and epistemic harms of AI sycophancy. AI and Ethics, 6(2), 168. https://doi.org/10.1007/s43681-026-01007-4
Venning, E., & Beech, D. (2026). Preparing for populism. Higher Education Policy Institute. https://www.hepi.ac.uk/reports/preparing-for-populism/
Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Harvard University Press.
Wright, A., Wilding, M., Lawler, M., & Lowe, M. (2025). Student working lives. Higher Education Policy Institute (HEPI). https://www.hepi.ac.uk/wp-content/uploads/2025/11/Student-Working-Lives.pdf
Xia, Q., Weng, X., Ouyang, F., Lin, T. J., & Chiu, T. K. F. (2024). A scoping review on how generative artificial intelligence transforms assessment in higher education. International Journal of Educational Technology in Higher Education, 21(40). https://doi.org/10.1186/s41239-024-00468-z
Yusuf, A., Pervin, N. & Román-González, M. (2024). Generative AI and the future of higher education: A threat to academic integrity or reformation? Evidence from multicultural perspectives. International Journal of Educational Technology in Higher Education, 21(21). https://doi.org/10.1186/s41239-024-00453-6
We are interested in your overall experience with your assessments so far in your course.
To what extent do you feel that the types of assessment you have taken have supported your learning overall? In what way(s)?
· Can you give some examples of the types of assessments you have found particularly beneficial in supporting your learning? In what ways have you found these beneficial?
· Are there any specific aspects of your assessments that have made it challenging (or a barrier) for you to engage in or to demonstrate your learning?
We would now like to focus on your perspectives and experiences in relation to diversity and choice in assessment methods.
What would you say are the benefits of having a variety of assessment methods as part of your course?
What do you feel would be challenging in having a variety of assessment methods as part of your course?
When completing assessments for your course, what are your thoughts about being given a choice in the format of the assessment?
· What are the benefits for you?
· What are the challenges?
We would now like to ask you to share your perspectives or experiences with any activities or resources which you feel have supported you in your assessments.
· Can you comment on any learning activities (e.g., workshops, resources, guidance) you have engaged with that you felt supported you in your assessment?
We are interested in your views on the use of Generative AI (Artificial intelligence) in the context of assessments.
· What opportunities, if any, have you had to engage with artificial intelligence (AI) to support your learning at university?
· Consider a context whereby in an assessment you are taking, the use of AI forms part of that assessment.
o What concerns would you have about using generative AI effectively in this assessment? What benefits do you anticipate this may have for you?
· What kind of support would be helpful for you to effectively use generative AI in your assessments, as per your course/module requirements or guidance?
Do you have any additional comments or suggestions regarding assessment practices?
|
Human-led learning in the Age of AI: From threat to learning partner |
Equity challenges in assessment practices |
Flexibility, diversity, and authenticity in inclusive assessment design |
|
Generative AI as a learning partner, supporting efficiency, understanding of content, revision. |
Variability in AI literacy among learners. |
Choice in assessment fosters inclusivity – flexibility in learning, supports diverse learner needs. |
|
Programming the AI tool to support critical, human-led use in learning. |
Availability of GenAI tools (free vs paid versions) can impact fairness in use. |
Choice in assessment supports autonomy and empowers learners. |
|
AI and academic integrity · ‘someone else’ seen as producing the work · use of AI is associated with cheating · questioning who is being assessed – the learner or the AI tool · some types of assessments seen as less meaningful in the age of AI · an appropriate approach to GenAI use should be adopted to preserve academic integrity and ownership of learning. |
Peer pressure – differential ways of using GenAI among learners can lead to an unfair advantage and unequal workload. |
Choice can support learner learning if this is appropriately embedded in curriculum design · ensuring constructive alignment · ensuring authentic assessment, aligned with professional practice · with opportunities for learners to develop a broad range of competencies. |
|
Use of AI associated with deskilling: loss of creativity, reduced learning, development of key discipline-specific competencies. |
Assessments that rely on memory (e.g., closed book, timed exams) are seen as unfair. |
Choice in assessment supports employability. |
|
Lack of trust in GenAI · AI can produce incorrect content · less reliable for some discipline-specific content · more trust if AI is purposefully built for the discipline. |
Detailed marking criteria and accompanying learning activities (e.g., formative assessment) supports assessment literacy.
|
Choice in assessment method can lead to uncertainty in relation to expectations. |
|
Guidance and support · Supporting the transition between forbidden to encouraged use of GenAI in assessment. · Supporting integration of GenAI in assessment – appropriate uses, AI literacy, critical use of AI. |
|
Diversity in assessment methods associated with concerns in effectively preparing for different types of assessments. |
|
|
|
Diversity in assessment can be linked to increased assessment load. |
|
|
|
Lack of diversity in assessment can lead to reduced breadth of learning. |