Page 267 - CyberSecurityEssentialsEnglish

6.6 Advising, Communication, and Student Support
◆ Key idea: support does not replace the human relationship.
AI can widen access to information and first-line support, but it does not replace the
human relationship and it does not take sensitive decisions about a learner. The best
design makes clear what the system can do, what it cannot, and when a request moves
to a responsible person.

Support for a learner extends beyond lesson time. A student or trainee needs to understand
requirements, organize time, find a service, interpret a comment, prepare for a meeting,

and choose a next step. Many of these questions recur, and AI can offer a fast first answer,
provided it rests on approved information and states its limits and its date.

A digital assistant can explain registration procedures, turn a long guide into steps, remind
people of deadlines, or help a learner break a project down. But the assistant must

distinguish general information from an individual case. A phrase such as “the form is usually
required before the deadline” is not the same as an official decision that a particular person’s

request is accepted or refused.

In academic advising, a tool can help a learner formulate goals, compare options, prepare
questions for an adviser, and build an initial plan. It should not decide alone on a course, a
specialization, or an academic future, because the decision may depend on recent policies,

prior requirements, personal circumstances, and goals that do not appear in the data.

Escalation must be designed from the outset. If a learner mentions a psychological crisis, a
threat, violence, discrimination, a health problem, a disciplinary decision, or a sensitive legal
or financial situation, the system must stop offering general advice and direct them to an

appropriate human channel according to the institution and the context. Gentle language is
not the same as professional competence.

Personalization in support can be useful, such as remembering a preferred language, a type
of appointment, or a plan the learner made. But it can turn into surveillance if a great deal

of data is gathered, if the system infers sensitive characteristics, or if it classifies learners
without their clear knowledge. Data minimization, transparency, the ability to correct, and a

rule against using an automated prediction alone for a high-impact decision must all apply.
It is also important that the assistant does not become a single compulsory gateway. Some
people cannot easily use a digital chat, and some need human contact or a different means
of access. Clear alternatives must be provided, a learner must not be penalized for choosing

another channel, and the system must be reviewed for whether it delivers comparable
quality across Arabic, its dialects, and different reading levels.

266