引導式測驗 · 硬 等級

AI 武器偵測掃描儀測驗

了解人工智慧篩選系統如何產生警報、為什麼供應商聲稱需要證據,以及學校如何評估檢測遺漏和誤報。

相關引導路徑AI 武器偵測掃描儀
問題 1 的 8

What does a weapon-screening alert indicate?

預覽每個問題8 問題

當您參加測驗時,就會出現答案和解釋。

  1. What does a weapon-screening alert indicate?
    • A sensor pattern needs secondary review
    • A weapon has been confirmed
    • The venue is fully secure
    • A person violated school policy
  2. Why evaluate false alarms alongside detection rates?
    • False alarms prove the system is working
    • Detection rates include every operating condition
    • Benign alerts affect staffing, access, and trust
    • False alarms cannot occur in AI systems
  3. What did the FTC allege in its Evolv action?
    • All scanner vendors use identical sensors
    • Claims about detection and harmless-item performance were misleading
    • Every school system committed a violation
    • AI screening systems are prohibited
  4. What should procurement teams request from vendors?
    • Only a statement that the product uses AI
    • A guarantee that no false alarm will occur
    • A testimonial from another venue rather than evidence from the offered model
    • Performance evidence for the tested version and conditions
  5. Why can a scanner miss an item?
    • Performance depends on configuration, environment, and tested objects
    • A sensor alert always finds every threat
    • Model training eliminates operational conditions
    • Operators cannot affect screening results
  6. What should happen after an alert?
    • Assume the alert proves a threat
    • Automatically deny all entry without review
    • Follow a trained secondary-screening procedure
    • Ask the AI system to make a disciplinary decision
  7. Why is local testing important?
    • Vendor demonstrations are illegal
    • The school’s conditions may differ from a vendor demonstration
    • Test results never transfer across systems
    • Local conditions determine the model’s architecture
  8. What does the presence of AI in a scanner prove?
    • That every weapon will be detected
    • That staff training is unnecessary
    • That privacy risks are absent
    • Only that AI is part of the product, not that performance is improved