Teaching · Practical guide
Make AI Use Transparent Without Guessing Authorship
Set task-specific AI boundaries, teach useful declarations and respond to concerns through evidence and a fair institutional process.
Define what the assessment is meant to demonstrate
Academic integrity begins with an intelligible task. If students do not know whether brainstorming, translation or feedback is permitted, a general instruction to “use AI responsibly” leaves important decisions unresolved. Identify the skill being assessed and explain which forms of assistance preserve or bypass that skill.
For a task assessing independent reasoning, generating the reasoning may be prohibited even when spelling support is allowed. For a task assessing the critique of an AI response, generated material is part of the evidence. Neither approach is automatically appropriate for every assessment. Follow the actual institutional policy and publish task-level instructions consistent with it.
Describe permissions as actions
Use concrete verbs: may brainstorm headings; may request feedback on an original draft; must verify sources; must not generate the final analysis. State the required declaration and where it belongs. Include a route for asking questions before submission and an equivalent pathway when students cannot access a named tool.
Avoid saying “AI is allowed” without limits. A student may interpret this as permission to generate a whole answer, while a teacher intended only proofreading. A short example of permitted and prohibited assistance often communicates more than a long abstract policy. Make clear that acknowledging a prohibited action does not make it permitted.
Worked example: an assessment-use statement
Fictional assignment: explain a networking calculation and justify two assumptions. Example task rule: “You may use an approved assistant to suggest practice questions and to flag unclear wording in your own draft. You must perform and explain the submitted calculation yourself. Do not submit generated reasoning as your own. Add a short declaration describing any assistance and your verification.” This is a sample design, not a claim about Bahrain Polytechnic policy.
Permitted illustrative declaration: “I used an approved assistant to flag two unclear sentences in my draft. I rewrote them myself and recalculated the percentages with a calculator. It did not produce my submitted analysis.” The statement is useful because it identifies the action, contribution and check. “AI was used” does not tell a marker what happened.
If a student instead asks an assistant to produce the entire calculation and explanation, copying it with a declaration still conflicts with this example rule. The response should follow the institution’s established procedure, including an opportunity for the learner to explain the work.
Distinguish a concern from evidence of misconduct
Unfamiliar vocabulary, polished writing or a sudden change in style can prompt a conversation, but they do not establish authorship. A writing-pattern percentage cannot show that a student used AI. DYAP’s Document Review tool explicitly reports rule matches, not an AI-authorship probability; it should not be treated as proof for grading or discipline.
Use relevant assessment evidence through the authorised process: ask the learner to explain a method, discuss a cited source or demonstrate a small equivalent task. Interpret that evidence in context, including the task instructions and permitted support. Do not secretly upload submissions to unapproved services or make a public accusation from a tool score.
Teach documentation before there is a problem
Offer a simple record with four fields: assistance requested, material used, changes accepted and checks performed. Students can retain relevant drafts or notes according to course requirements. Do not demand complete private account histories or unrelated personal conversations as a routine condition of learning.
A process record is supporting evidence, not a guarantee. It can be incomplete, and a polished log can be fabricated. Combine reasonable documentation with well-designed opportunities to demonstrate learning. TEQSA’s assessment reform resources provide a broader institutional discussion; this guide’s sample wording is an original practical exercise rather than a universal policy.
Practice: rewrite an ambiguous rule in 15 minutes
Take the fictional instruction “AI may be used responsibly”. Spend four minutes identifying the assessed skill. Use five minutes to write two permitted actions, two prohibited actions and a declaration requirement. Spend six minutes testing the wording against brainstorming, proofreading, generated analysis and citation checking.
Example answer for a source-evaluation task: allow search-term brainstorming and language feedback; prohibit inventing references and substituting an AI summary for reading the required sources. Require a declaration of assistance and a claim-to-source evidence table. The learner must still judge whether each source supports the claim.
Self-review: would two students interpret the boundaries similarly? Is the permitted assistance consistent with the skill being assessed? Does the procedure for concerns rely on relevant evidence and institutional rules rather than a detector score? Record one unresolved policy question and seek the appropriate institutional answer before using the rule in a real assessment.
Reusable prompt
Select and copy this template, or customise your own using the free builder.
Review this fictional assessment rule for ambiguity: [rule]. Assessed skill: [skill]. Identify what is permitted, prohibited and unclear. Propose concrete examples and an assistance-declaration template. Do not infer misconduct, diagnose authorship or invent institutional policy. Flag decisions requiring the course owner’s approval and preserve a fair opportunity for learners to explain their work.Open the prompt builder →
Pause · Think · Check
Test your understanding
Choose one answer per question, then check your reasoning. This is practice; scores are not saved.