Page 29 - CyberSecurityEssentialsEnglish
P. 29

FIGURE 1-2: APPROACHES TO BUILDING AI SYSTEMS AND HOW THEY RELATE

         What to notice: Machine learning is part of AI, and deep learning is part of machine learning.

         Most modern generative AI applications rely on deep learning, although other methods of
         generating content also exist.
         Source: Prepared by the author based on references [3–5].



                   ✕ Common misconception                                        Correction


        Every  program  that  runs  automatically  is  AI,  Some  programs  use  fixed  automation.
        and  every  AI  system  learns  from  data  and  Some AI systems are symbolic or hybrid.
        thinks like a person.                                  Performing  an  intelligent  task  does  not

                                                               demonstrate  consciousness  or  human
                                                               understanding.


         What Is Not AI in Itself?

         1. Database: A database stores and retrieves information. An AI system may use it, but the
         database does not learn or generate an answer by itself.
         2. Search engine: A search engine finds and ranks pages or documents. A generative system

         creates new content, although some applications combine search and generation.
         3. Automation: Automation may execute fixed steps without learning. AI can be part of an
         automated process, but the two terms are not synonymous.

         4. Consciousness: Fluent conversation or simulated emotion does not prove that a system
         feels or intends what it says.

         5. Authoritative source: A confidently worded answer does not become an official record. It
         must be checked against the relevant source and context.





         28
   24   25   26   27   28   29   30   31   32   33   34