Use generative AI with clear academic boundaries: follow assessment rules, verify claims, protect personal data and keep a record of assistance.
Generative AI can produce fluent explanations that still contain mistakes or invented references. Its value in study depends on the task, the course rules and your ability to evaluate the output. A useful starting principle is to remain responsible for every claim and every source in work submitted under your name.
Check permission before choosing the tool
Read the assessment instructions and ask when permitted use is unclear. Brainstorming questions, getting feedback on structure and generating submitted answers are different activities; a course may treat them differently. Do not assume that a tool's availability means its use is allowed in every assessment.
Build a verification step into the workflow
For a suggested fact, find an independent primary source and read it. For a citation, confirm that the work exists and supports the statement. Test an explanation against a worked example. Keep a short log of prompts, useful suggestions and corrections so you can describe the assistance accurately if disclosure is required.
Protect learning and confidentiality
Avoid entering personal records, unpublished work belonging to others or confidential organisational data without appropriate permission. After using a tool, explain the idea without it and solve a related problem yourself. If you cannot do that, the polished output has concealed a learning gap that still needs attention.
Worked example: verify a suggested reference
Imagine that a tool supplies a reference for comparing two study methods. Search the publisher’s website or library catalogue using the author and title, then read the relevant passage if the work exists. It might address different learners or reach a different conclusion. Keep only what you can verify and develop your interpretation from the actual source. If the reference cannot be located, remove it instead of asking the tool to make it sound more plausible.
Make the assistance explainable
Where assistance is permitted, note the tool, date and task, such as suggesting a structure for draft headings. Record which suggestions you accepted or changed and why. This does not replace the required disclosure format, but helps you complete it accurately. Before submission, choose a paragraph and explain its reasoning and sources without opening the conversation. If that is difficult, return to reading and independent work before relying on the generated material in your answer.
Further reading
- UNESCO — Guidance for generative AI in education and research
- The Open University Library — Referencing and plagiarism
Image: original editorial artwork prepared for the Asas library; it illustrates the topic and does not depict an actual event or student.
