Page 263 - CyberSecurityEssentialsEnglish

6.5 Helping the Learner Without Doing the Work for Them

◆ Key idea: helping does not mean solving.
Good educational help reduces the obstacle, not the necessary mental effort. It begins by
understanding where the learner is stuck, then offers the least sufficient support, and
withdraws gradually until the learner takes back responsibility for the solution and the
explanation.

When a learner faces a hard task, asking for the complete solution is the fastest route to

finishing the work, but it is not always the fastest route to learning. The difference between
support and substitution shows in a simple question: after the interaction, is the learner

more able to carry out a similar task alone? If the answer is no, we have moved the product
from the tool to the page without building capability. The literature on the relationship
between learners and generative AI stresses treating the tool as a limited support partner

rather than a complete cognitive substitute [18].

Temporary support that helps a learner accomplish what they cannot accomplish alone is
called scaffolding. It should be tied to the need and be removable. It may begin by clarifying

the requirement, breaking the task down, recalling a rule, asking a question that directs
attention, or offering a similar example. The complete solution is the highest degree of help,
and it is not used automatically at the start.

Before offering help, the system or the teacher needs a diagnosis. What did the learner try
to do? Where did they stop? How do they interpret the problem? Is the gap in the concept,

the language, the procedure, or time management? If the learner says “I do not understand
anything”, the tool can ask two or three questions to locate the starting point instead of
delivering a comprehensive explanation that may increase the confusion.

An attempt-first protocol is useful in many tasks. The learner writes what they know, an

initial plan, or a prediction, and then asks for specific support. After receiving help, they
revise their attempt and explain what changed and why. This simple trace makes thinking

visible and reduces the risk of the tool becoming a hidden substitute.
AI can be directed to act as a coach rather than a writer. For example: “Do not give the final
answer. Ask me first about my attempt, then give one hint, and wait for my reply. If I am

wrong, explain the type of error and give me a similar problem.” This wording does not
guarantee perfect behavior, but it establishes a better educational relationship than a
request for the solution.

Learners also need metacognitive skills: planning, monitoring, and evaluating. Before the
task they ask: what is required, and what is my plan? During it: am I understanding or
repeating? And after it: what worked, and what will I change? AI can ask reflective questions,

262