引導式測驗 · 中等 等級

自動化政府決策程序測驗

測試自動公共決策的正當程序:受保護的利益、馬修斯因素、通知、證據存取、人工審查、上訴和審計記錄。

相關引導路徑自動化政府決策程序
問題 1 的 8

Under Mathews v. Eldridge, which set of factors guides procedural due-process analysis?

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. Under Mathews v. Eldridge, which set of factors guides procedural due-process analysis?
    • Model accuracy, vendor cost, and number of users
    • Whether a decision used artificial intelligence
    • The agency’s preference for automation
    • The private interest, risk of error and value of safeguards, and government interests
  2. An agency notice says only “algorithmic score too low.” What is missing?
    • A longer model name
    • Case-specific reasons and enough information to challenge the underlying facts
    • A second score from the vendor
    • A guarantee of success on appeal
  3. What makes human review meaningful?
    • The reviewer can inspect source evidence, consider contrary information, and change an incorrect result
    • The reviewer clicks approve on every score
    • The reviewer sees only the final number
    • The vendor certifies the system once
  4. Why can’t due-process requirements be reduced to one universal checklist?
    • Every government decision has identical consequences
    • Algorithms are outside due-process law
    • The protected interest, governing law, decision stage, and consequences vary by context
    • An agency may select any process it prefers
  5. Which log helps reconstruct an automated denial?
    • Only the final score
    • The vendor’s marketing page
    • A total number of cases processed
    • Source inputs, data source, software and rule version, reviewer action, notice, and appeal history
  6. An appeal reviewer cannot see the records that drove a score. What problem follows?
    • The score becomes more reliable
    • The person’s appeal is automatically resolved
    • The reviewer may be unable to assess the accuracy or relevance of the decision
    • No source review is needed if the model is consistent
  7. Which measure tests whether people can understand an adverse notice?
    • Number of characters in the model output
    • Comprehension testing with affected users, including accessible formats and language needs
    • Average time to compute a score
    • Count of source-code comments
  8. When should an agency re-evaluate its decision process?
    • When data, thresholds, models, or decision consequences materially change
    • Only when a vendor contract expires
    • Only after a court orders it
    • Whenever the model produces a high score