AI has become a permanent feature of higher education. Across the country, lecturers are increasingly concerned about students using AI tools such as ChatGPT, Gemini and Claude to complete assignments.
In response, universities have invested in AI detection software while developing policies, guidelines and frameworks aimed at promoting ethical AI usage.
Yet not all institutions are convinced that detection software offers a sustainable solution. Questions are increasingly being raised about the reliability of the systems, their potential for false positives and whether they address the root cause of the problem. Put differently, can universities detect their way out of the AI challenge?
Having recently completed the first semester of the 2026 academic year, I found myself reflecting on the question. Across several assignments, in one module I was teaching, one issue stood out. A significant proportion of assignments triggered high levels of AI detection. Like many academics, my initial reaction was concern. Were students taking shortcuts? Has AI become the latest threat to academic integrity?
The more I reflected on the situation, however, the more I began to wonder whether we, as universities, were asking the wrong question. Instead of focusing solely on why students are using AI, perhaps we should also be asking why our assessments make AI use so attractive in the first place? If large numbers of students are turning to AI to complete assignments, then the issue might not lie solely with student behaviour. It might also reveal something about the kinds of assignments we set, the skills we choose to assess and the assumptions we hold about learning in an AI-enabled world.
The possibility raises an uncomfortable but necessary question: Are universities partly responsible for the AI behaviours they are trying to prevent?
If universities are serious about responding to AI, then institutional readiness must go beyond policy documents and detection tools. It requires practical capability development, particularly among lecturers who remain central to the design of teaching, learning and assessment. As higher education adapts to an AI-enabled world, there are five questions that lecturers, like myself, should be asking.
First, do we need assignments that primarily require students to reproduce information?
For decades, university assignments have been built around assessing independent research, information synthesis, critical thinking and academic writing. Traditionally, the tasks required students to spend considerable time locating information, evaluating sources and constructing arguments.
Generative AI has changed this landscape. Today, a student can enter an assignment question into an AI system and receive a coherent response in seconds. Tasks that once required hours of work can be completed almost instantly. The issue is not whether students are using AI. They are. The real question is whether our assessments are measuring what we think they are measuring.
As lecturers, we need to ask ourselves: Are we genuinely assessing learning or are we assessing information retrieval? When students submit assignments that can be generated by AI with minimal effort, what exactly are we rewarding? Are we evaluating intellectual development or are we merely satisfying a compliance requirement that demonstrates a student has submitted a piece of academic writing? The question is no longer whether AI can complete assignments. It can and we know that. The more important question is whether our assignments measure meaningful learning.
Second, are our assignments connected to real-world problems?
Recently, I attended an event bringing together academics and practitioners. The contrast in attitudes towards AI was striking. Many academics repeated the familiar warning to students: "If I catch you using AI." Practitioners, however, expressed frustration that graduates entering the workplace were often hesitant to use AI to support routine tasks.
The disconnect reveals a deeper issue. Universities often claim to prepare students for the world of work and active citizenship. Yet many assignments remain theoretical and disconnected from the realities students will face beyond graduation. Students are frequently asked to define concepts, compare theories and summarise literature. While such activities have value, they often occur within an artificial academic environment where demonstrating knowledge becomes more important than applying it.
Generative AI exposes the limitations of the approach because it performs exceptionally well at reproducing theoretical content. It can explain concepts, compare theories and summarise articles in seconds. The more abstract and generic the task, the easier it becomes for AI to generate an acceptable response.
Universities therefore need assessments that encourage critical and independent thinking while remaining connected to real-world challenges. Students should be asked to engage with realities, solve contextual problems and apply disciplinary knowledge to authentic situations.
Imagine a public administration student developing solutions to service delivery challenges in a municipality they know. Consider a business management student evaluating the sustainability of a local enterprise. Or a health sciences student analysing community health interventions in a familiar context.
Such assessments demand more than information retrieval. They require contextual understanding, judgement and engagement with real problems. Importantly, AI can play a role. Students might use it to brainstorm ideas or gather information. The difference is that AI becomes a resource among many rather than the source of the final answer.
Third, who is doing the thinking?
Another priority is ensuring that lecturers and students remain firmly in control of the educational process, with AI serving as a tool rather than a substitute for learning.
Much of the public discussion about AI is dominated by extremes. Some predict that AI will replace large parts of teaching and assessment. Others fear that students will lose the ability to think independently.
Both positions overlook a simple reality: higher education is a human endeavour.
Universities are not merely institutions that transfer information. They are places where students learn to question assumptions, engage with diverse perspectives, develop professional identities and cultivate intellectual curiosity. The outcomes emerge through dialogue, mentorship, feedback and human interaction.
AI undoubtedly offers benefits. It can assist with brainstorming, provide writing support and automate routine tasks. For lecturers, it can reduce administrative burdens. For students, it can offer personalised assistance.
However, AI should augment human capability rather than diminish it.
Fourth, are we doing enough to develop AI literacy among students and lecturers?
There is a danger that universities mistake access to AI tools for genuine AI literacy. Too often, AI literacy is treated as a technical skill focused on prompts and outputs. In reality, it is also an ethical and critical capability. Students need to understand not only what AI can do but also where it can fail.
Many students encounter AI as a source of answers rather than as an object of critical enquiry. They learn how to use it but not necessarily how to question it. This is problematic because AI-generated outputs can appear authoritative even when they are inaccurate.
Lecturers therefore have a crucial role to play. Students need practical exposure to the strengths and weaknesses of AI systems. They should learn how AI can perpetuate bias, generate hallucinations, fabricate references and reproduce societal inequalities embedded in training data. They should also understand that AI systems can produce outputs that appear confident despite being wrong. Such examples help students develop healthy scepticism and critical judgement.
Finally, should universities move from surveillance to stewardship?
Perhaps the greatest danger in the AI debate is that universities become so preoccupied with detecting AI use that they neglect their educational mission. Across the world, institutional responses to AI have focused on detection, monitoring and punishment. While academic integrity remains important, a surveillance-centred approach risks creating cultures of mistrust.
Students become reluctant to discuss AI use. Lecturers become focused on identifying violations rather than facilitating learning. AI use is pushed underground rather than engaged with openly.
Universities should consider a different approach: stewardship.
Stewardship recognises that higher education institutions have a responsibility not merely to regulate AI but to guide its ethical use. The question is not whether students are using AI. They are. The more important question is whether universities are helping them use it responsibly, transparently and critically.
A stewardship approach shifts the conversation from punishment to education. It encourages students to disclose how AI was used, explain how outputs were evaluated and reflect on how technology contributed to learning.
Ultimately, the future of higher education will not be determined by whether AI enters our classrooms. That has happened. Nor will it be determined by how effectively universities detect AI-generated text. The future will depend on whether institutions are willing to rethink assessment, literacy, ethics and institutional culture for an AI-enabled world.
The challenge before higher education is therefore not technological. It is educational. The task is not simply to police AI but to steward its use in ways that strengthen learning, preserve human agency and prepare graduates for an increasingly AI-mediated world.