Staff Guide: Assessing Student Generative-AI Use
A practical guide for teaching staff on reaching fair, consistent decisions when student AI use is suspected or disclosed.
Demonstration writing sample. This guide illustrates my professional and instructional writing. Adapt it to your institution’s actual policies before operational use.
Purpose
This guide helps you move from a suspicion or disclosure of generative-AI use to a decision that is fair, consistent and explainable to the student.
Before you begin
- Locate the rules that actually apply: the institutional integrity policy, the unit outline and the task instructions. The most specific instruction the student was given governs the case.
- Write down what the student was told was permitted. If you cannot state it in one sentence, treat ambiguity as a factor in the student’s favour.
The six questions
Work through these in order and record your answer to each:
- Was AI use permitted? Quote the applicable instruction.
- Was it disclosed? Check the declaration against what the evidence shows.
- Did AI assist or replace the student’s work? Identify where the substantive intellectual contribution came from.
- Did the student verify the output? Look for evidence the student checked accuracy, references and reasoning.
- Was authorship preserved? Can the student explain and defend the work?
- Was there deception? Distinguish an incomplete declaration from an actively false account.
Reaching a decision
- All six favourable → acceptable use.
- Permitted, authored, verified, but not declared → disclosure problem; respond to the declaration failure, not as contract cheating.
- AI replaced the work, or the student cannot account for it → potential misconduct; follow the formal procedure.
- The rules do not decide the case → record it as unclear and escalate the policy gap, not the student.
What not to rely on
- AI-detection scores are not evidence of authorship. Treat them, at most, as a prompt to look at the work itself.
- Fluency, vocabulary or “sounding like AI” is not evidence.
Talking to the student
Put the specific concern to the student, show them the rule involved, and let them respond before forming a view. An authorship conversation should probe understanding of the submitted work — supportively, and with allowance for anxiety and language background.