Page 351 - CyberSecurityEssentialsEnglish
P. 351

extraction, the synthesis, the verification, and the documentation.
           • Practical decision: Complete one small auditable package rather than a long text built
           on disorganized searching.
           • Mistake to avoid: Beginning to write the report before the chain of evidence has been
           built.


         Chapter Conclusion: The Chain of Evidence Matters More Than the Speed of the Text
         The story of the policy draft withdrawn because of fabricated references reminds us that the
         greater danger is not that AI makes mistakes, but that it gives a mistake an official form that

         is hard to doubt. An organized text is not evidence, a written reference is not a source until
         its existence and its fitness are established, and a fast summary does not become knowledge
         until it returns to an origin that can be inspected.


         AI can make research faster and better organized: it widens the search vocabulary, organizes

         files,  helps  extract  fields,  writes  initial  code,  proposes  comparisons,  and  improves  the
         language of a report. But every benefit becomes a risk if it is severed from the source, the

         method, and the responsibility. Speed that produces traceable evidence is valuable; speed
         that produces confidence without verification is a deferred cost.

         Professional  practice  begins  with  a  clear  question,  then  a  documented  search  strategy,
         screening based on criteria, reading that ties findings to their locations, a synthesis that

         preserves difference and uncertainty, writing that links claim to evidence, verification of
         references, and disclosure proportionate to the tool’s effect. These practices do not belong
         to universities alone; they belong to every report, policy, or decision that claims to rest on

         knowledge.

         The human being remains responsible for choosing the question, protecting participants,
         understanding  the  method,  interpreting  the  results,  weighing  values  and  risks,  and
         defending the claims. In the next chapter the book moves from the responsibility of the

         researcher  and  the  team  to  the  level  of  the  organization:  how  are  policy,  governance,
         accountability, and the safe adoption of AI built inside an institution?

           “An organized text is not evidence. A reference is not a source until it exists, matches, and
           carries the sentence attributed to it.”














        350
   346   347   348   349   350   351   352   353   354   355   356