Google Kuis AI

Jawaban dan penjelasan muncul saat Anda mengikuti kuis.

1. Which distinction does the guide make about Gemini?
   • Gemini models, the Gemini application, and developer tools have different roles
   • All Gemini services expose identical features
   • A model name guarantees every tool is enabled
   • The consumer app is the same thing as a cloud deployment
2. A feature appears in the Gemini consumer app. What should a developer do before relying on it through an API?
   • Assume the same feature is exposed in the API
   • Rely on an old product announcement
   • Check the relevant API documentation and project availability
   • Use the feature name in a prompt
3. Why should model and feature availability be checked for the intended project?
   • All projects share identical quotas
   • Quotas and access can vary by model, endpoint, account, or region
   • Availability depends only on how many prompts are written
   • A consumer app’s feature list controls every API
4. What should a developer review before sending private material to a Google AI service?
   • Only whether the project uses the largest model
   • Assume a consumer app’s privacy settings apply to every developer project
   • Assume no request data is retained whenever training use is disabled
   • Review data terms and retention controls for the chosen API service
5. A report image contains a small footnote. What should a developer account for when using a multimodal model?
   • The model may miss fine text; input quality and resolution can affect what it reads
   • The API verifies every number against the original report automatically
   • The image-processing step guarantees every visible word becomes a model input
   • Repeating the same prompt ensures that all small text will be recognized
6. Does retrieval or search support remove the need to verify an answer?
   • Yes; retrieval guarantees every claim
   • Only if the source is official
   • No; check the final claims against the sources
   • Yes, if the answer includes links
7. What does a model capability establish about a specific app workflow?
   • That the feature is automatically enabled everywhere
   • Only potential capability; account permissions and enabled features also matter
   • That the complete workflow has been tested
   • That data controls are identical
8. What should be recorded when evaluating a Google AI configuration?
   • The company name without its model or endpoint
   • Only the final answer produced in one successful run
   • The project name without the selected model and settings
   • The model, project settings, inputs, and evaluation date