The purpose of this viewpoint is to contemplate AI (artificial intelligence) in assessment within a higher education setting from the position of an academic with limited AI expertise. The piece is to prompt thought, discussion and action within a highly contemporary debate.
The approach is UK academic based using a reflexive/autoethnographic form of discussion based on the author’s 20+ years of higher education experience. Crucially, the position presented is one of limited AI expertise, which is hoped to offer accessibility to the topic and suggestions for likeminded academics.
The viewpoint presents a somewhat neutral perspective of AI in higher education assessment from an experienced academic with limited AI expertise. Recommendations are offered to those akin to this position: (1) stay in touch with AI news and changes relating to assessment approaches and potential impacts within higher education, (2) voice concerns/questions and become an active learner and (3) develop new AI skills. This advice is offered within a context of difficult current challenges within higher education and therefore a need to be more proactive in such times.
A viewpoint like this offers the possibility to discuss relevant issues not only in a timely fashion but also in a highly accessible way. It brings together some key insights from a fairly neutral perspective so that academics can tap into key discussion points to take forward into their own actions.
I have been told by quite a few academics that it is “when” not “if” artificial intelligence (AI) will become a normative practice for higher education assessment marking. Indeed, a recent HEPI (2025) survey highlights an incredible 88% of students using GenAI for assessments in 2025 – a staggering increase from 53% in 2024. How should any academic feel about that? I am certainly at the lower end of knowledge when it comes to AI, so I argue it is worth presenting this somewhat skewed viewpoint to enable wider access to the topic. The hope is it sparks thought and discussion using a reflective academic voice with limited AI expertise. Those with knowledge and power can serve to teach the rest of us about what big important aspects we are missing or lacking. This paper is not delving into the discussion around AI and ethics (and the broader debate around this), for example a paper summed up perfectly within its title of “The wicked problem of AI and assessment” (Corbin et al., 2025) and Selwyn (2019) exploring discussions around values, judgements and politics. It is a crucial discussion, but one outside of the parameters of this reflective piece. However, this paper is still an important contribution to responsible enterprise pedagogy through a focus on AI and assessment.
For those more akin to my knowledge-limited position, it may just prompt us to get involved in the discussion in the first place (rather than minimising our interaction). The second half of the main title is a “tongue-in-cheek” remark to represent an academic like myself that has considerable higher education knowledge and experience (the pants), but substantially lacking in terms of AI knowledge (the lack of trousers). The statement perhaps reflects where many of us are, that is an expectation from the industry that we are knowledgeable and prepared, but a crucial AI knowledge deficit persists (that we may or may not want to admit out loud).
To contemplate my perspective living and working in UK higher education, the recent national picture has been a bleak one for many. Within this, academics are facing greater job uncertainty/ insecurity than ever before, increased workload pressures (especially regarding teaching) and extensive institutional constraints. Indeed, there is a constant drive for cost saving, efficiency and then innovation towards the next big successful programme to conquer the competitors. It is then not a great surprise that there is great interest in what innovation and savings AI could bring in the medium-to-long term for programmes everywhere. Traditional face-to-face delivery renowned for expense may suddenly come with dramatic savings to help push those limited incomes from student fees so much further than before.
Our perspective as academics on the frontline could be a double-edged sword. On the one hand, the constant drag of endless marking could be substantially reduced leading to other possibilities (e.g. greater time for pedagogical (re)design, enhanced student support or more room for reflective teaching practice) – potentially sounds great (see Dr Deepshikha’s work via Advance HE, 2025, for a modern example). But on the other hand, it surely takes away important workload hours we have maintained through our careers, that is we are not needed as much. A classic AI dilemma – the replacement of skills. An optimist might say “Yes, more time for research … (and other valuable activities)”, but the pessimist might say “Oh no, that means even more teaching to be done”, that is the dread of further labour intensification in these challenging times.
I am left with only one certainty at this point of writing – the AI juggernaut does not appear to be slowing down. I feel change will come sooner rather than later regarding AI assessment practice, but I sense it will be steps rather than “overnight” systemic change. This viewpoint paper presents a reflexive position on the topic of AI and assessment. In essence, the discussion points have an autoethnographic feel to them, yet the paper is not a full research study – it retains the essence of a viewpoint discussion.
What might we expect from AI in assessment?
Assessment redesign: Whether we are embracing AI capability or preparing to mitigate issues around AI usage (e.g. students cheating), it appears we are already in the process of assessment redesign. There is substantial and perhaps convincing evidence out there for adopting AI in assessment for students, teachers and institutions (e.g. Xia et al., 2024). Yet, I have seen a number of strategies in practice concerning mitigation of potential issues we seem ill equipped for in higher education, including (1) a potential return to written exams and (2) many more oral styles of assessment. In essence, these appear to be AI avoidance strategies attempting to sever the link to AI adoption for students at the critical point in the assessment construction and delivery. Not sure these are the solutions to the overarching issues, but these approaches can perhaps serve to slow down a need to change or could be used in conjunction with AI adoption, that is maybe a good way to enhance academic integrity (or at least provide greater confidence of integrity).
Enhanced personalised feedback: This appears to be a big one as AI can produce extensive and individualised feedback in the blink of an eye where it takes a human considerably more time. Again, there will be questions over integrity and relevance, but the prospect of depth and speed seems an opportunity that cannot be missed, especially when thinking back to the higher education context and drive to cost save. Yet, perhaps for now the expectation will be much more about combining human judgement with AI output. This will inevitably come with a need for new skills and learning for academics.
Enhanced learning support: This is the one topic perhaps most in the limelight currently, yet it feels like a taboo topic. There is a lot of discussion (e.g. Zhao, 2024, provides a wide scoping systematic review of pros and cons), policies, split opinions and so on around adoption of AI, especially for assessments. There is a great fear that students will turn to AI to “cheat”. I use that phrasing intentionally, as, like many new things we experience, negative assumptions are often made without proper evidence. Yes, AI could be used to “cheat”, but it is clearly argued that it can also be used as an effective learning support tool. It presents a real dichotomy in terms of what to do from an academic perspective. Thus, this perhaps emphasises the importance of earlier points around assessment redesign and finding strategies to ensure academic integrity.
Potential recommendations for academics with limited AI expertise
Academics need to stay in touch with AI news and changes relating to assessment approaches and potential impacts within higher education. Indeed, we have varying practices already regarding AI as part of student assessment submission process, and universities are working on addressing that as well as trialling new AI marking processes. It may sound obvious, but I am reminded of change when we shifted from hand written assessments to typed/electronic, the shift from submitting a print out “at the office” to electronic submissions, the introduction of Turnitin software (including the much criticised AI reporting tool in more recent times) and the full integration of virtual learning environments, whereby assessment guidance and support can be delivered (i.e. not just for submissions). Those who dragged their feet in terms of learning these new ways often came up against issues and barriers. I actually remember many academics retiring seeing this landscape changing their traditionalist ways of delivering teaching and assessment. What might happen to the next generation of academics faced with this profound landscape change? Is there research needed to perhaps assess these perspectives?
Voice concerns/questions and become an active learner. Who do we speak to about these changes to come? The network of potential knowledge is very important, and we all need those “clued up” connections to help advance our knowledge and thinking.
Develop new AI skills. For those like me, this is the scary one; stepping into the abyss to fully embrace a world alien to us. Yet, it seems we must get ahead of the curve, or at least much closer to it. I fear my skills will become disassociated with those at the cutting edge of higher education. Alas, I finish with a “call to arms”: Let us be brave and, despite not having the time or finding 100 reasons not to do it, let us interact and interface with AI software and become better thought leaders on the subject of AI in assessment.
The future
For AI in assessment, it is clear there is a need for much more ongoing research. Here are a few suggestions for potential avenues that I feel would benefit a wider AI knowledge-limited readership:
What about a Special Issue on AI in assessment?
How about researching multi-stakeholder perspectives, for example students, academics, parents, programme leaders/directors, module/unit leaders, admin staff, accrediting bodies and so on?
How about case examples of trying AI in assessment approaches?
What about perspectives globally and exploring similarities and differences?
In conclusion, I am hoping this viewpoint reflects my current perspective – that is neither particularly pro- nor anti-AI. Yet, so much discussion sits on one side of the fence or the other. I hope a more neutral view adds something to the discussions we are having, that is by having a lack of AI expertise, I am not driving a particular AI agenda one way or another. Ultimately, I am a realist and I can see AI continuing to grow, evolve and integrate into the world of higher education. I sense I need to be more proactive in my learning and approaches (within these extensive wider challenges, we currently face as academics), and I suspect others need to as well. If anything, that is to ensure I preserve my relevance and skills as an academic with a lot more years to come before retirement.
