Page 271 - CyberSecurityEssentialsEnglish

● Remember: Section summary
• Key idea: Automated support suits information, orientation, and triage, not the
relationship or the sensitive decision.
• Practical decision: Tie the assistant to approved sources, an escalation path, and an
available human being.
• Mistake to avoid: Measuring success by the number of conversations, or letting the

system decide an individual case.

6.7 Academic Integrity in Teaching and Learning
◆ Key idea: integrity means showing the learning.
Academic integrity does not mean preventing all help. It means that the work reflects the

learning of the person who submitted it, that the help was permitted and declared, and
that the assessor is not misled about who did the thinking, the writing, or the
performance.

AI has opened a wide space between traditional individual work and clear cheating. A learner

may use a tool to understand instructions, generate revision questions, improve language,
propose a structure, or write the whole answer. The judgment does not rest on the presence
of the tool alone, but on the purpose of the task, its rules, the size of the contribution, and

whether the use conceals an absence of learning.

Every task needs a clear rule at the task level, not only a vague general policy. Using AI may
be required in an activity that critiques its outputs, permitted with conditions in a draft, and
prohibited in an examination that measures individual performance. When a learner knows

the limits of use and the method of disclosure before starting, the grey area shrinks and
fairness increases. Sector guidance on generative AI in higher education offers practical

examples of the value of clear rules and prior disclosure [17].

It helps to distinguish assistance, co-creation, substitution, and misrepresentation.
Assistance removes a barrier or offers feedback while the learner remains the owner of the
idea and the work. Co-creation integrates the tool’s contributions in a permitted and

declared way. Substitution happens when the tool performs the essence of the task.
Misrepresentation happens when a person submits work that does not reflect their
contribution, or breaks the rules and conceals it.

The only response should not be a search for “AI-proof” assignments. Some general take-

home tasks can be generated easily however their wording changes. The deeper reform is
to diversify the evidence of learning: drafts, discussions, local applications, live performance,
oral explanation, a decision log, and a commentary explaining how the tool was used and

what was rejected.

270