Page 48 - CyberSecurityEssentialsEnglish

Stage Human role System role

Makes the decision, remains
Does not replace the accountable
Approval accountable, and documents
person or professional authority.
important changes.

▣ Practical example: A university professor reviews teaching material
The system can suggest explanations, examples, and questions. The professor defines the
learning objective, checks the concept and level of difficulty, and decides what is
appropriate for the students. The value lies in expanding the available options and
reducing preparation time, not in transferring responsibility for teaching to the model.

● Remember: Section summary
• Key idea: Human judgment extends from defining the objective to approving the
decision.
• Practical decision: Create a genuine review point led by someone with the necessary
knowledge and authority to reject the output.
• Mistake to avoid: Treating a person as a final approval stamp after the system has
effectively defined the goal, interpreted the context, and made the decision.

1.10 Practical Lab: Mapping AI Use in Your Work

◆ Key idea: Begin with a small experiment whose results you can

measure.
Do not start with a large transformation project. Choose a real, low-risk task, divide it into
steps, define AI’s role, prepare test examples, and compare time, quality, and errors. The
purpose is to discover where the technology adds genuine value to your work.

Step One: Choose One Task
Choose a task that you repeat, that takes a noticeable amount of time, and whose output

you can evaluate. Examples include summarizing a report, organizing meeting minutes,
extracting fields from forms, turning material into questions, or comparing three

alternatives. For your first experiment, avoid sensitive data and high-impact decisions.

Step Two: Analyze the Task Before the Tool
1. What is the input: text, a table, an image, or a recording?

2. What output is required: a category, number, ranking, draft, plan, or recommendation?
3. Is the appropriate role acceleration, augmentation, or automation?

4. What is the impact of an error, and who can detect it?
5. What authoritative source or standard will you use for comparison?

47