Викторина по подкастингу с использованием искусственного интеллекта

Ответы и пояснения появляются по мере прохождения теста.

1. Сценарий подкаста обобщает исследование. Что продюсер должен сохранить в заметках к шоу перед выпуском?
   • A generated paraphrase without a source record
   • Only the model’s confidence score for its summary
   • Links to the primary sources used to verify the spoken claims
   • A list of prompts in place of checking the study
2. Когда сгенерированная цитата может быть приписана реальному собеседнику?
   • When evidence confirms that the person said those words
   • When the synthetic voice resembles the person’s voice
   • When the generated quote sounds plausible in context
   • When a transcript model assigns a high confidence score
3. Устный черновик содержит плотные цифры и необъяснимые сокращения. Какая редакция улучшает четкость прослушивания?
   • Raise the audio level without changing the script
   • Move every qualification to a link the listener may not open
   • Read the abbreviations faster to shorten the episode
   • Explain abbreviations and simplify comparisons while retaining important qualifications
4. Синтетического рассказчика можно было принять за оригинальную запись настоящего гостя. Что должно произойти перед публикацией?
   • Hide the guest’s name in the transcript
   • Obtain relevant voice permission and disclose the synthesis to listeners
   • Change the audio codec so the voice sounds less synthetic
   • Disclose the edit only if the guest objects afterward
5. Почему резюме исследовательской работы в подкасте, созданное с помощью искусственного интеллекта, необходимо проверять перед публикацией?
   • Comparing only the narration with the generated script verifies the paper’s findings
   • The summary may misstate findings or invent details absent from the source
   • A citation in show notes proves that every spoken claim matches the paper
   • Agreement among other generated summaries replaces checking the paper
6. Before using a voice clone of a guest to fix a line, what is most important?
   • Get explicit permission for the synthesized line and disclose it when listeners may mistake it for the guest’s original recording
   • Treat permission to record the interview as permission to synthesize new words
   • Put the synthesis disclosure only in private production notes
   • Rely on a generic recording release that never mentions synthetic speech
7. An editor removes a segment after an automatic transcript was made. What needs checking before release?
   • Only the transcript wording, assuming its timestamps and references still match
   • Only the audio cut, assuming the transcript and summary update automatically
   • Only the show-note links, assuming the transcript still follows the old edit
   • Transcript timing, speaker labels, show-note references, and summaries affected by the cut
8. The episode upload succeeds. What still checks the listener’s final experience?
   • Assume the host preview matches the delivered file
   • Play the exported episode and compare its transcript and show-note links with the final audio
   • Count words in the draft script
   • Inspect only the transcription model’s confidence