HDMI went out for a family lunch on Sunday. It was one of those long, lingering lunches, sitting in the shade of the trees while conversation meandered from one topic to another.
As people swapped seats over the course of the meal, I found myself drawn several times into conversations about AI, academic integrity, and plagiarism. All hot topics at the moment.
Do students cheat? Are they allowed to use AI? Are universities still teaching subjects that will be needed in the future?
You can imagine the conversation.
Over a number of years, and across different institutions, I have worked with academic communities to co-create approaches to academic integrity and, more recently, to the use of generative AI. My starting point has consistently been that we need to teach our students how to use the tools available to them with academic integrity.
That means being clear about what tools they have used, how they have used them, and what contribution they have made to their work.
A student might use a generative AI tool to help with some initial desk-based research, to explore a topic, or to identify avenues they subsequently investigate themselves. What they should not do is pass off AI-generated output as their own work.
The same principle applies to academics. We may use AI tools to support elements of our work. If we do, we should be transparent about that too. Academic integrity cannot be something we demand of students while quietly adopting a different standard for ourselves.
AI is going to be part of all our futures. It is therefore incumbent upon universities to ensure that both staff and students develop the skills and competencies needed to use it well.
That raises a more difficult question: What if some of the things we currently teach, and some of the things we currently assess, can now be done perfectly adequately by generative AI – bearing in mind that it will only ever get better, and rapidly?
The answer cannot simply be to prohibit the technology. In some cases, it may be telling us something important about our curriculum and its design.
We need to move beyond asking whether students should be ‘allowed’ to use AI, as Sam Illingworth has explored in his HEPI study of how UK university AI policies actually work. As I have argued elsewhere, providing both AI-free and AI-supported assessment, is likely to become normal practice.
Some assessments should be deliberately AI-free, because we need to know what students can do independently. Others should be deliberately AI-supported, because we need to know whether students can use these tools critically, ethically, and effectively.
What matters is that the purpose is clear. Academic integrity then becomes less about policing the presence of a tool, and more about whether a student has used it in ways that are appropriate, transparent, and accountable. Sam’s report points to something similar: a shift from assessing tool proficiency towards the assessment of critical literacy.
Assessment design is one part of the answer. The curriculum itself is another.
Courses need continually to evolve. If what we are teaching has not kept pace with technological and societal change, parts of the curriculum may increasingly be overtaken by what contemporary AI tools can produce.
That does not make Higher Education redundant. Quite the opposite I would say.
Higher Education has always been about more than acquiring knowledge in a particular subject. It is also about developing the capacities that accompany that knowledge: learning how to think, how to question, how to work with others, how to evaluate evidence, how to communicate, and how to reach a judgement.
Those capabilities become more important, not less, when information and apparently plausible answers are available instantly.
Former Google CEO Eric Schmidt put it starkly:
"If you're writing code in any traditional way: stop. It's over."
He says he is “in mourning” as a programmer, having spent decades doing work that AI can now do faster. Whatever you make of the framing, it puts real weight behind a question Higher Education cannot avoid.
The shape of some disciplines may change too. We should be asking ourselves questions such as:
- Given that AI agents can write code, and will only get better at doing so, how should we reflect that in our Computer Science or Software Engineering curricula?
- In the future, which forms of digital capability will society need when machines and tools can undertake more of the coding themselves?
- Within ‘computing’ how could we place greater emphasis on students learning how to apply digital approaches to healthcare, engineering, the creative industries, public services, and problems we have not yet encountered?
This will require portfolio review and curriculum renewal. Universities need to educate students not simply for the jobs or skills priorities identified by a government today, but for lives and careers that will unfold rapidly over the years to come.
It also means teaching students, and ourselves, how to work knowledgeably with AI:
- How do we prompt an AI agent well?
- How do we interrogate what comes back?
- How do we recognise hallucination, bias, or a superficially convincing answer?
- How do we decide when AI is the right tool, and when it is not?
- Perhaps most importantly, how do we remain accountable for the final output?
For me, that is where academic integrity sits.
We live in an information-rich world. I see nothing inherently wrong with drawing upon the information, knowledge, expertise, and tools that society has created in producing something new. The important issue is whether we are honest about where ideas and contributions have come from.
- If someone else has said something better than I can, I quote them and cite them.
- If someone else can do something better than I can, I learn from and collaborate with them.
- I, like most people these days, use a calculator, instead of doing maths in my head.
- If I cannot remember who said precisely what about a subject, I, like you I am sure, ‘Google it’, find the original source, read and check it, and cite it.
Using a tool, in itself, does not show a lack of integrity. Pretending that work, ideas, or outputs derived from elsewhere are entirely our own, is. Perhaps that is the distinction we need to hold on to as AI becomes increasingly ordinary.
Academic integrity should not require us to pretend that we work alone, unaided by the extraordinary body of knowledge and technology around us. It should require us to exercise judgement over what we use, to remain accountable for what we produce, and to be transparent about how we have worked.
Colleagues Dr Hazel Farrell and Ken McCarthy make this same case directly in their Manifesto for Generative AI in Higher Education: transparency is the new integrity. I could not agree more.
For transparency, this piece was conceived, drafted, and written by me. I then used an AI tool to help edit and refine it.
Further Reading
Sam Illingworth, What UK university AI policies actually do: A study of 96 institutions, Policy Note 71, HEPI https://www.hepi.ac.uk/reports/what-uk-university-ai-policies-actually-do-a-study-of-96-institutions
Eric Schmidt, If you’re writing code traditionally, stop. It’s over, Economic Times https://economictimes.indiatimes.com/news/new-updates/if-youre-writing-code-traditionally-stop-its-over-former-google-ceo-eric-schmidt-explains-why-he-is-mourning-as-a-programmer/articleshow/131364339.cms
Hazel Farrell and Ken McCarthy, Manifesto for Generative AI in Higher Education, GenAI:N3 https://manifesto.genain3.ie
Harriet Dunbar-Morris, AI, assessment and belonging, HEPI Blog https://www.hepi.ac.uk/2026/05/01/ai-assessment-and-belonging
