Page 159 - CyberSecurityEssentialsEnglish

+ Methodological note: 4D is a teaching aid, not an official standard
The 4D framework used in this chapter is a teaching simplification prepared by the
author for organizing everyday use. The official NIST AI Risk Management Framework is
organized around functions such as govern, map, measure, and manage, and the
generative AI profile applies that view to generative systems. We use 4D as a practical
bridge, not as a substitute for official frameworks or for an organization’s own policies.

TABLE 4-1: THE DIFFERENCE BETWEEN A LINGUISTICALLY GOOD OUTPUT AND A PROFESSIONALLY GOOD USE

The decisive
A linguistically good output A professionally good use
question

Is the problem Organized text that answers A clear task, with limits on what
defined? the request. the tool should do.

Is the data It processes everything sent It uses the minimum permitted

appropriate? to it. data.

Is the result Coherent and confident Claims tied to evidence and
correct? wording. verified.

It produces a suggestion or an Human approval, a stopping
Can it be reversed?
action. point, and a correction plan exist.

Do we know what A record of sources, version,
Only the final output is visible.
happened? changes, and the decision exists.

✕ Common misconception Correction

Assess the evidence, the data, the impact,
If the text is convincing and free of and the permissions. Fluency is a linguistic

language errors, it is safe to use. property, not a certificate of correctness
or safety.

● Remember: Section summary
• Key idea: A good prompt raises quality but does not settle privacy, the impact of error,
or the correctness of sources.
• Practical decision: Evaluate the whole use, not the final text alone.
• Mistake to avoid: Equating a professional appearance with correct information or an

appropriate decision.

158