Idanwo ti o sopọ mọ itọsọna · Alabọde Ipele

AI ni Imudaniloju Ofin ati Awọn adanwo ọlọpa

Review policing AI uses and limits, including face-match thresholds, feedback from arrest data, gunshot alerts, and official legal classifications.

Jẹmọ itọsọna awọn ọnaAI Ni Imudaniloju Ofin Ati ọlọpa
Ibeere 1 ti 8

What did NIST face-recognition evaluations report about demographic error-rate differences?

Awotẹlẹ gbogbo ibeere8 ibeere

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

  1. What did NIST face-recognition evaluations report about demographic error-rate differences?
    • Most tested algorithms showed demographic differentials, with size depending on algorithm and application
    • All tested algorithms had identical error rates for every demographic group
    • The study found demographic differences only in license-plate readers
    • NIST concluded that image quality never affects error rates
  2. Kini idi ti awọn alariwisi sọ pe ọlọpa asọtẹlẹ le ṣe agbega abosi?
    • Ko lo data rara
    • O kọ ẹkọ lati inu data imuni ti o ṣe afihan iṣẹ ọlọpa ti o kọja
    • O ṣe asọtẹlẹ oju ojo nikan
    • O foju pa ipo patapata
  3. Kini eto idanimọ oju ṣẹda lati aworan oju kan?
    • Afọwọya iwe kan
    • A itẹka
    • Ifisinu nọmba tabi 'oju oju'
    • A DNA ọkọọkan
  4. ShotSpotter jẹ apẹẹrẹ ti iru eto?
    • Oluka iwe-aṣẹ
    • Idanimọ oju
    • DNA onínọmbà
    • Akositiki ibon erin
  5. Awọn oluka iwe-aṣẹ aladaaṣe ni akọkọ:
    • Wọle awọn ipo ọkọ ni iwọn
    • Tumọ awọn ede ajeji
    • Ṣe iwọn didara afẹfẹ
    • Ipinfunni iwe irinna
  6. Bawo ni Ofin EU AI ṣe tọju ọpọlọpọ imufinfin AI ti nlo?
    • Bi ko ṣe ilana
    • Bi eewu giga tabi eewọ
    • Bi dandan fun gbogbo olopa
    • Bi Idanilaraya software
  7. In a face-recognition system, what can changing the match-similarity threshold trade off?
    • Image resolution against video length
    • Database size against audio quality
    • False matches against missed matches
    • Encryption strength against network speed
  8. What can AI transcription add to a body-worn-camera reporting workflow?
    • A determination of guilt without a court
    • A facial-recognition match from audio alone
    • An automatic guarantee that a report is accurate
    • A searchable transcript or draft narrative based on recorded audio