Page 279 - CyberSecurityEssentialsEnglish

a model. But the alternatives must measure the same outcome. If the objective is academic
writing, it cannot always be replaced by a video; but if the objective is to explain a concept,
several fair formats may exist.

Plain language is an important access tool. Use clear sentences, define terms, break

instructions apart, highlight the required action, and avoid padding. AI can produce a
simplified version, but the two versions must be compared to make sure no condition,

constraint, or specialist meaning has been removed.

In bilingual education, a tool can help prepare a glossary, give an Arabic definition alongside
the English term, and explain the difference between a literal translation and specialist
usage. Names, symbols, and text direction must be tested, and it must not be assumed that

performance is equal across languages and dialects. The learner keeps the right to use their
language as a bridge to understanding rather than as an obstacle.

Assistive technologies such as screen readers, speech recognition, and alternative input
devices need well-structured files. Real heading styles, a logical reading order, table labels,

alternative text for images, captions for video, and appropriate contrast are all required. AI
can propose the alternative text or the captions, but technical and human testing remain
necessary.

Neurodiversity and differences in attention, organization, and processing must also be

considered. Step lists, flexible timers, task decomposition, and options that reduce
distraction may help. But a system should not diagnose a condition or decide on support

without an institutional process and the learner’s participation. The best support is built with
the person, not about them.

Inclusion is incomplete without thinking about connectivity, devices, and cost. An activity
that requires a paid model or a permanent connection may exclude some people. A low-

technology alternative should be available, the core materials should be accessible without
complicated accounts, files should be tested on a phone, and the quality of learning should
not depend on a personal subscription.

Representation and fairness inside the examples must also be examined. Do the

occupations, roles, abilities, and identities appear in varied and non-stereotyped ways? Do
the examples reflect local contexts? Does the language respect people? A tool can be used
for a first audit, but the review must involve real people and experiences rather than being

left to a model that learned from historical patterns.

The practical rule is: do not assume that conversion equals accessibility. Turning text into
audio does not guarantee its clarity, generating a translation does not guarantee its

278