引導式測驗 · 中等 等級
選舉和公民技術測驗中的人工智慧
探索記錄在案的選舉人工智慧風險、媒體出處限制以及基於模擬的地區地圖分析方法。
問題 1 的 8
2024 年 1 月新罕布夏州初選中發生了什麼凸顯了人工智慧選舉風險?
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。
- 2024 年 1 月新罕布夏州初選中發生了什麼凸顯了人工智慧選舉風險?
- 人工智慧生成的機器人電話模仿拜登總統的聲音,告訴選民不要投票
- 投票機被人工智慧入侵
- 人工智慧聊天機器人給了錯誤的投票地點
- Deepfake選票已郵寄給選民
- C2PA 標準的目的是什麼?
- 加密選民登記資料庫
- 在捕獲時對媒體進行加密簽名以驗證來源
- 更快地計票
- 封鎖所有網路政治廣告
- Under California’s vote-by-mail signature rule, what happens when a signature has multiple significant and obvious differences from the voter’s record?
- It is automatically accepted without review
- It is automatically rejected in every case
- It receives additional review by an elections official
- It is replaced by an AI-generated signature
- What training-example duration does the cited Neural Voice Cloning paper say can suffice in some few-shot setups?
- At least several weeks of continuous audio
- Exactly one hour in every model
- No speaker audio is ever used
- A few minutes or even a few seconds, depending on the setup
- 如何利用人工智慧來偵測黨派不公正劃分選區?
- 透過模擬數千個可能的地區地圖進行比較
- 透過手動計票
- 通過禁止重新劃分選區
- 透過翻譯選票
- In its 2024 TCPA declaratory ruling, how did the FCC classify AI-generated voices used in covered calls?
- As certified proof that a political claim is true
- As artificial or prerecorded voices under the TCPA
- As a voter-registration record
- As a voice that is exempt from all call-consent rules
- What does a valid C2PA Content Credential primarily communicate?
- A universal verdict that every statement in the media is true
- A guarantee that the media has never been edited
- Claims about media provenance and its recorded history
- Proof that a human rather than software created every element
- In the cited gerrymandering study, how are districting plans assessed?
- By asking a language model whether a map looks fair
- By checking only the party label of the mapmaker
- By guaranteeing one mathematically perfect district map
- By comparing a proposed plan with an ensemble of simulated plans under stated assumptions