Quiz collegato alla guida · Medio Livello

Quiz: come l'intelligenza artificiale deduce tratti sensibili che non hai mai condiviso

Distinguere i tratti dedotti dai fatti, esaminare come i dati ordinari possono rivelare informazioni sensibili e mappare le protezioni della privacy in base ai dati e all'utilizzo.

Domanda 1 di 8

Which description best captures an AI inference about a person?

Anteprima di ogni domanda8 domande

Le risposte e le spiegazioni vengono visualizzate mentre rispondi al quiz.

  1. Which description best captures an AI inference about a person?
    • A verified statement the person made
    • An estimate derived from other information
    • A public court ruling
    • A sensor’s serial number
  2. What did the 2013 Facebook Likes study show?
    • Every user’s health diagnosis is known
    • Some private traits could be predicted from recorded behavior in the study dataset
    • Likes never reveal information
    • The study measured GPS only
  3. Does a model’s prediction that someone is depressed establish a diagnosis?
    • Yes, if confidence is high
    • Only if the app collected location
    • No; it is a prediction, not a clinical diagnosis
    • Yes, if the user clicked a health page
  4. Which example can generate an inference without a direct profile field?
    • Repeated location visits combined with other patterns
    • A blank database
    • A user’s explicit signed statement only
    • An unrelated screen color
  5. Which data types are treated as special categories under the GDPR?
    • Only online purchases
    • Health and political-opinion data, among other listed categories
    • All website cookies automatically
    • Only public information
  6. Can a sensitive inference be wrong even if the model is sophisticated?
    • No, models verify hidden traits
    • Only if the data are encrypted
    • Yes; it can be uncertain, biased, or overconfident
    • No, because it uses many inputs
  7. What should an organization document about an inferred trait?
    • Observed inputs, inference method, confidence, purpose, and recipients
    • Only the model’s marketing name
    • Only the final prediction
    • Nothing if the person never shared it
  8. Why should a prediction not silently become an eligibility rule?
    • It may be an uncertain proxy not necessary for the product purpose
    • All eligibility rules are prohibited
    • Predictions cannot be stored
    • Users prefer hidden decisions