Page 258 - CyberSecurityEssentialsEnglish

6.4 Quizzes, Rubrics, Feedback, and Spaced Learning
◆ Key idea: assessment can build learning.
Assessment is not limited to issuing a grade at the end of learning. It becomes a learning
tool when it reveals the gap early, pushes the learner to retrieve and explain, offers a
clear next step, and then allows another attempt.

AI helps produce a large number of questions, but quantity does not mean quality of
measurement. A good question is tied to a defined learning outcome, has a defensible

answer or criterion, uses appropriate language, and contains no unintended cues. If it is a
multiple-choice question, the wrong options must be plausible and reflect common errors,
not comic statements that can be eliminated at once.

We distinguish formative from summative assessment. Formative assessment happens

during learning and aims to guide the next step; it may be an oral question, a draft, or a short
ungraded quiz. Summative assessment issues a judgment at the end of a unit or a course. AI
can support both, but the risks and the required human oversight are higher in summative

assessment. This is consistent with the view of formative assessment as a tool for improving
learning while it happens rather than merely issuing a final judgment [13].

Different cognitive levels can be used when designing questions: remember, understand,
apply, analyze, evaluate, and create. These levels are not a rigid ladder, but their value is

that they prevent a course from testing the memory of terminology while declaring that its
goal is problem-solving or decision-making. Ask the tool to distribute questions across

defined mental operations, then check what each question actually requires.

Retrieval practice means attempting to recall knowledge from memory rather than simply
rereading. Learning research has shown that retrieval itself can strengthen long-term
retention, especially when it is followed by correction [7][14]. AI can create repeated short

questions, but it must not display the answer before the learner has tried, and it must not
turn the activity into a race for speed alone.

Spaced practice distributes review across time rather than massing it into one session. It can
be combined with interleaving, mixing related types of problems so that the person learns

to choose the appropriate strategy rather than executing a predictable procedure. A tool
can prepare a review schedule, select items that need revisiting, and generate variations,
but the load, the time, and the alignment with the curriculum must be monitored. The

literature on effective learning supports combining retrieval with spacing over simply
rereading [14].

Feedback is not a phrase such as “excellent work” or “try again”. It must connect the goal,
the current evidence, and the next step. A well-known model summarizes three questions:

257