Kvíz spojený s průvodcem · Střední úroveň

Lidský dohled podle článku 14 zákona EU o umělé inteligenci

Zkontrolujte, zda rozumíte lidskému dohledu podle článku 14 aktu eu, jeho rozsahu, klíčových odpovědnostech a praktických omezeních v rámci EU.

Otázka 1 z 8

Which systems are the direct focus of Article 14 oversight requirements?

Náhled každé otázky8 otázky

Odpovědi a vysvětlení se zobrazují, jak budete v kvízu.

  1. Which systems are the direct focus of Article 14 oversight requirements?
    • High-risk AI systems
    • Every software feature that uses an algorithm
    • Only general-purpose AI models
    • Only AI systems used by EU institutions
  2. Under Article 14, human oversight aims to do what?
    • Guarantee that every output is correct
    • Transfer all legal responsibility to a reviewer
    • Prevent or reduce risks to health, safety, or fundamental rights
    • Require manual approval of every output
  3. Which practical condition makes oversight meaningful?
    • A person is listed as the supervisor but cannot access outputs
    • The assigned person can understand, challenge, and intervene in the system’s use
    • All outputs are automatically accepted after a quick click
    • The system provides no way to pause or override
  4. In this context, automation bias means what?
    • A model changing its parameters during training
    • A reviewer refusing to use any digital tools
    • A legal presumption that every AI output is biased
    • Over-relying on an AI output because it is machine-generated
  5. Which is an oversight capability named in Article 14?
    • Changing the system’s training data during deployment
    • Approving every output without inspection
    • Deciding not to use the system output or to override it
    • Removing the provider’s technical documentation
  6. How should oversight measures be selected?
    • In proportion to risk, autonomy, and the context of use
    • Use the same checklist for every system regardless of use
    • Base them only on the vendor’s marketing claims
    • Choose the least expensive option in all cases
  7. Does a human reviewer alone satisfy all high-risk AI Act duties?
    • Yes, human presence replaces technical requirements
    • Yes, if the reviewer signs an annual form
    • No, because the Act prohibits human oversight
    • No; oversight is one requirement alongside other applicable safeguards
  8. Who must assign appropriately qualified people to oversee use?
    • Only the person affected by an AI-assisted decision
    • The deployer, within its operational responsibilities
    • The AI system itself
    • The European Parliament for every system