Page 24 - CyberSecurityEssentialsEnglish

1.1 AI Around Us: Where Does It Add Value?

◆ Key idea: We already use AI in many everyday tools.
Spam filtering, route suggestions, search-result ranking, recommendations, speech
recognition, and unusual-transaction detection are all examples of AI systems.
Conversational tools have made AI more visible, but they are neither the beginning of the
field nor its only form.

AI is not limited to tools that write text or answer questions. When a map application
estimates arrival time, a bank flags an unusual transaction, a phone converts speech to text,

or an online store ranks products, the system processes data to produce an output for a
specific task. That output may be a forecast, ranking, classification, text, or suggested action.
At work, value begins with identifying where time is being lost, not with choosing the most

popular tool. Much of the working day is spent collecting notes, organizing text, extracting
data, comparing versions, summarizing information, and reformatting material. If a system
completes these preliminary tasks in a form that can be reviewed, people can devote more

time to work that requires expertise and judgment.

Breaking a task into steps helps reveal where AI can provide real value. “Preparing a report”

may look like a single activity, but it includes collecting and organizing data, calculating
indicators, deciding on the message, drafting the text, checking the figures, and formatting
the result. AI may suit some of these steps but not others. If it accelerates drafting but

introduces numerical errors, it has not saved time; it has merely shifted that time to review.
We should therefore measure the full workflow and the number of corrections, not just the
speed of text generation.

AI’s contribution can be understood through three roles. Acceleration reduces the time

required for a known step, such as drafting or summarizing a document. Augmentation helps
people analyze more options or compare several alternatives. Automation carries out
repeated steps according to defined rules, permissions, and stopping points. The more work
we delegate to a system, the more testing and monitoring it requires.

23