Quiz połączony z przewodnikiem · Średni Poziom

Holenderski quiz dotyczący algorytmów świadczeń opieki nad dziećmi

Testuje, w jaki sposób holenderski model ryzyka wykorzystywał narodowość, dlaczego szkoda wynikła z całego systemu, a nie tylko algorytmu i powiązania wysokiego ryzyka z ustawą o sztucznej inteligencji.

Powiązane ścieżki prowadząceSkandal w sprawie holenderskiego algorytmu świadczeń opieki nad dziećmi
Pytanie 1 z 8

Which factor did the tax administration's risk classification model use that was found to be discriminatory?

Podgląd każdego pytania8 pytania

Odpowiedzi i wyjaśnienia pojawiają się w miarę rozwiązywania quizu.

  1. Which factor did the tax administration's risk classification model use that was found to be discriminatory?
    • The parent's income bracket
    • Non-Dutch nationality
    • The number of children in daycare
    • The childcare agency's size
  2. What did the Dutch Data Protection Authority conclude in 2020?
    • The model was lawful but poorly documented
    • Only the childcare agencies had broken the law
    • The algorithm had never been used
    • The processing of nationality and dual nationality data was unlawful and discriminatory
  3. What happened to parents labeled as acting with intent or gross negligence?
    • They were refused payment arrangements for their debts
    • They received extra allowance
    • Their files were automatically closed
    • They were given priority appeals
  4. What was the title of the December 2020 parliamentary inquiry report?
    • Xenophobic Machines
    • Odchylenie maszynowe
    • Unprecedented Injustice
    • Fraud Signaling Facility
  5. What political consequence followed the scandal?
    • The tax administration was abolished
    • Prime Minister Mark Rutte's cabinet resigned in January 2021
    • The Netherlands left the EU AI Act negotiations
    • Childcare allowances were eliminated
  6. How can a feedback loop make a fraud risk model confirm its own bias?
    • Models automatically delete old data
    • Human reviewers always overrule the model
    • Random sampling removes all bias
    • Flagged groups get more scrutiny, more discrepancies are found, and those findings become training labels
  7. According to the guide, what actually caused the harm in the scandal?
    • The algorithm alone
    • The risk model combined with harsh recovery rules, a presumption of fraud, secrecy and weak appeals
    • Only the use of nationality, which would have been fixed by removing it
    • Fraud by the childcare agencies themselves
  8. How does the EU AI Act classify AI used to assess eligibility for public benefits?
    • As minimal risk with no obligations
    • As fully prohibited in all cases
    • As high risk
    • As outside the Act's scope