- Rose Stephenson is Director of Policy and Advocacy at HEPI.
On the eve of the release of HEPI’s Student Generative AI Survey 2025, HEPI hosted a roundtable dinner with the report’s sponsor, Kortext, and invited guests to discuss the following essay question:
How will AI change the university experience for the next generation?
This was the third roundtable discussion we have hosted with Kortext on AI, over three years. Observing the debate mature from a cautious, risk-averse response to this forward-looking, employability-focused discussion has been fascinating. We spent much of the evening discussing a potential pivot for teaching and learning in the sector.
The higher education sector places the highest importance on creating, collecting, and applying knowledge. ‘Traditional’ assessments have focused on the recollection of knowledge (exams) or the organisation and communication of knowledge (in essays). The advent of search engines has made acquiring knowledge more accessible, while generative AI has automated the communication of knowledge.
If knowledge is easily accessible, explainable, and digestible, which skills should our graduates possess that cannot be replaced by ChatGPT, now or in the future? It was suggested that these are distinctly ‘human’ skills: relationship building, in-person communication, and leadership. Are we explicitly teaching these skills within the curriculum? Are we assessing them? Are we rebalancing our taught programmes from knowledge to irreplaceable skills to stay ahead of the AI curve?
And to get a bit meta about it all, what AI skills are we teaching? Not just the practical skills of application of AI use in one’s field, but deep AI literacy. Recognising bias, verifying accuracy, understanding intellectual property rights and embracing digital ambition. (Professor Sarah Jones of Southampton Solent University has written about this here.)
Given recent geopolitical events, critical thinking was also emphasized. When and why can something be considered the ‘truth’? What is ‘truth’, and why is it important?
Colleagues were clear that developing students’ knowledge and understanding should still be a key part of the higher education process (after all, you can’t apply knowledge if you don’t have a basic level of it). In addition, they suggested that we need to be clearer with students about the experiential benefits of learning. As one colleague stated,
‘The value of the essay is not the words you have put on the page, it is the processes you go through in getting the words to the page. How do you select your information? How do you structure your argument more clearly? How do you choose the right words to convince your reader of your point?’
There was further discussion about the importance of experiential learning, even within traditional frameworks. Do we clearly explain to students the benefits of learning experiences – such as essay writing – and how this will develop their personal and employability skills? One of the participants mentioned that they were bribing their son not to complete his Maths homework by using ChatGPT. As students increasingly find their time constrained due to paid work and caring responsibilities, how can we convince students of the value of fully engaging with their learning experiences and assessments when ChatGPT is such an attractive option? How explicitly are we talking to students about their skills development?
There was a sense of urgency to the discussion. One colleague described this as a critical juncture, a ‘one-time opportunity’ to make bold choices about developing our programmes to be future-focused. This will ensure graduates leave higher education with the skills expected and needed by their employers, which will outlast the rapidly evolving world of generative AI and ensure the sector remains relevant in a world of bite-sized, video-based learning and increasing automation.
Kortext is a HEPI partner.
Founded in 2013, Kortext is the UK’s leading student experience and engagement expert, pioneering digitally enhanced teaching and learning in the higher education community. Kortext supports institutions in boosting student engagement and driving outcomes with our AI-powered, cutting-edge content discovery and study products, market-leading learner analytics, and streamlined workflows for higher education. For more information, please visit: kortext.com
The issue here is GenAI resilient (rather than GenAI resistant) education. The knowledge reproduction may still be expected in early levels in Higher education, hence traditional assessment formats (such as invigilated exams, short answer questions with time restrictions etc.) may still be needed. However, once the knowledge accusation is ascertained, it is the “transformation of that knowledge into skills” (that is application) is important. This is where we can make GenAI as students’ buddy to work together for assessment by ensuring the assessment is focusing on human input in the form of verification, amalgamation (of GenAI generated pieces) in context of application via critical thinking etc.
Most importantly, we need to provide clear guidance to the students on what is expected. I have made three different titles to provide this guideline.
1. GenAI unaided Assessments (AI limited for revision only)
2. AI assisted assessments
3. GenAI engaged assessments
Although many collogues argue the GenAI unaided assessments would become redundant over the year, my argument is as long as there is a need to asses the understanding of the knowledge (in the form of “reproduction”) there is a need to have this guideline to let the student know they cannot use GenAI during the assessment per se.
Of yes the GenAI is definitely going to shift the ways the learning and its applications, more towards the latter. In my experience I come across two types of academics (a) those who are extremely critical about GenAI and (b) those who are extremely engaging with GenAI. It is the balance that is needed.
This discussion underscores a crucial turning point for higher education. AI isn’t just shifting the balance between knowledge and skills—it’s demanding a complete disruption of how we define learning itself. If generative AI can automate knowledge production, then why are we still assessing students as if they are competing with machines?
Traditional assessment models—essays, exams, and rote learning—are rapidly becoming obsolete. Instead of merely rebalancing curricula, universities must radically redesign learning experiences to center on critical inquiry, problem-solving, and lived experience. This isn’t just about preparing students for jobs; it’s about equipping them to challenge, question, and reshape the world AI is creating.
I have long advocated and worked in the spaces that interconnect skills, qualifications, experience, attributes and knowledge (SQuEAK, if you like acronyms) — and articulated them in terms of the ‘behavioural competencies’ that evidence and demonstrate these elements in learning, work and life in general. All intersect and are important — students can integrate and personalise their learning with the holistic development of essential transferable skills and attributes (behaviours and actions). These include critical and creative thinking, appreciating diversity and being adaptive in collaborating with others and making major collective decisions, information literacy (important in relation to GAI and fake news, etc.). All embedded in my ‘SOARing to Success’ pedagogy and andragogy, with tools that are available in my publications. I would be more than happy to work with Kortext as you might see fit. This approach has benefited thousands of students so we know it works!