“Use artificial intelligence responsibly” seems straightforward until a student is alone with an assignment brief, a blank document and a deadline. For many international students, particularly those on intensive one-year master’s programmes, the challenge extends beyond subject knowledge. They must interpret academic English, navigate UK higher education assessment conventions and understand institutional AI rules. Broad policies are necessary, but they rarely help students make sound decisions when it comes to assessments. They need clarity about what AI can support, what they must produce independently and how to distinguish between the two.
We have been exploring ways to offer this in our Level 7 module for MSc engineering management students, many of whom come from the Global South. Assessment is built around a practical design problem: students examine how a real campus process operates, where delays occur, evidence they observe and how a feasible redesign might improve the system – in a 3,000-word essay.
They can use AI for planning and providing clarity, but not for generating the evidence on which the analysis rests. Invented observations, measurements, screenshots or unverifiable citations are prohibited. The assessment design reinforces this through a case study, diagram, fact-checking exercise and a short AI use statement.
Address uncertainty at the outset
Students often ask the following questions: Can AI explain the brief? Can it clarify “critical analysis”? Can it help with English language expressions? Can it suggest sources?
To support, we ask students to paste part of the assignment brief into a GenAI tool and request a plain English explanation, then to compare it with the official brief. They must consider what the tool clarifies and which requirements need to be checked against the marking criteria.
This approach positions AI use as part of assessment literacy while highlighting the authority of the official brief and marking criteria.
- Spotlight guide: Critical thinking in teaching and research
- Spotlight guide: AI and assessment in higher education
- Can academics tell the difference between AI-generated and human-authored content?
Make the boundary clear
Students need clear examples rather than general warnings about “misuse”. In this assessment, the boundary is explicit: students may ask AI to explain where a process slows down, suggest questions for analysing a campus process or improve the clarity of a paragraph they have written.
They may not ask AI to invent observations, create fictional survey data, produce a fake screenshot, generate a diagram of a process they did not examine or supply unchecked references. If students count how many people are waiting as a result of a slow process, that queue count is evidence and must come from their own observations, not from AI.
They must also include at least one diagram and one reconciliation check each. A reconciliation check is a basic consistency check between parts of the evidence – for example, whether an observed queue length is plausible when compared with the waiting time estimate.
The principle is to design the task so that students must bring evidence from the real learning context that AI cannot legitimately supply.
Teach students to use AI around the task
Once students know which parts of the task must come from their own observations, evidence and judgement, they can use AI more confidently around those parts of the work. They can ask: “How can I structure a short case analysis?” “What theory might help me understand waiting time?” and “Can you make this paragraph clearer without adding new claims?” These prompts support comprehension and revision, particularly for multilingual students, but do not replace students’ evidence.
Many are experienced learners; the challenge is adapting to a new assessment culture at a time when AI tools are reshaping independent study. A clear assessment design reduces guesswork and prevents false confidence that can arise from polished AI-generated text. Students still need to justify their claims and demonstrate the source of the evidence.
Adapting to a new assessment landscape
Across the sector, debates remain about whether written assessments are fit for purpose. Some advocate for more verbal assessments, workshops or exams to protect academic integrity. These approaches have value but are not universally applicable. Over‑reliance on oral presentations risks disadvantaging students who speak English as an additional language, particularly without sufficient formative support. Simultaneously, many real-world tasks still require written outputs, often produced using AI tools. The aim is not to remove written assessments but to redesign them so that higher-order thinking, process evidence and judgement carry the most weight.
Institutions are exploring ways to allow students to use AI in assessment, with many awarding higher marks when they demonstrate critical thinking. This approach has potential but requires clear guidelines and consistent messaging. Staff views on AI vary, which can confuse students. Cohesive assessment design across courses is essential.
Ask for a short AI use statement
Requiring an AI use statement encourages students to reflect on their processes. We ask which tool they used, how they used it, how they checked the output and which parts of the work depended on their own evidence and judgement. It reminds them that evidence must be theirs and that citations must be manually verified.
Responsible AI use cannot rely on slogans. It must be designed into the work we ask students to do.
Andrew Firr is course leader for the MSc in engineering management at the University of Chester. Alex Fenton is associate dean for international for the Faculty of Science, Business and Enterprise at the University of Chester.
If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.
comment